* Add net.jqwik:jqwik dependency for property based testing
* Add maven-surefire-plugin config to include surefire-junit-platform so jqwik test are picked up and ran during `mvn test`
* Add new profile `idea-jqwik` that will add junit-platform associated libraries on a conditional basis so that IntelliJ Idea will detect jqwik tests as unit tests.

Co-authored-by: BenWhitehead <BenWhitehead@users.noreply.github.com>

…entation Http specific in HttpStorageOptions

1. Make StorageOptions and its Builder abstract classes. According to clirr this is a breaking change, however both classes had private constructors affording us this grey area to change things.
2. Add HttpStorageOptions which contains the http specific stuff from StorageOptions.
3. Deprecate the public inner classes of StorageOptions in favor of the HttpStorageOptions counterparts.
4. Refactor StorageImpl to always use HttpStorageOptions
5. Update all tests to use HttpStorageOptions
6. Remove unnecessary downloadTo tests from Blob now that the implementation is delegated to Storage
7. CopyWriter is still hard-coupled to HttpStorageRpc, this will need to be broken at a later point

*Compatibility Note*: This factoring is client API and ANI compatible,
however, due to the structural changes Serializable objects will not be
loadable from previous versions. serialVersionUID's have been updated
for the new class structures.

* Add new grpc implementation of Storage - GrpcStorageImpl
* Add new method StorageOptions#http() for an http specific builder
* Refactor internal usages of StorageOptions#newBuilder() to StorageOptions#http()

Everything is skeleton right now, but should allow for relatively quick implementation from multiple work streams.

For reasons I've not yet been able to root-cause, our dependency check builds are failing with the following error:
```
~/work/java-storage/java-storage/google-cloud-storage ~/work/java-storage/java-storage
2022-05-09 23:27:50 Generating dependency list using original pom...
2022-05-09 23:27:57 Comparing dependency lists...
2022-05-09 23:27:57 Diff found. See below:
2022-05-09 23:27:57 You can also check .diff.txt file located in google-cloud-storage.
34a35
> [INFO]    com.google.re2j:re2j:jar:1.5:runtime
43a45
> [INFO]    io.grpc:grpc-services:jar:1.45.1:runtime
47a50
> [INFO]    io.opencensus:opencensus-proto:jar:0.2.0:runtime
49a53,54
> [INFO]    org.bouncycastle:bcpkix-jdk15on:jar:1.67:runtime
> [INFO]    org.bouncycastle:bcprov-jdk15on:jar:1.67:runtime
~/work/java-storage/java-storage
```

It seems that something with runtime -> runtime -> runtime dependencies isn't properly showing up in the list of dependencies. In order to force them to show up, I've explicitly added a dependency on io.grpc:grpc-xds.

This explicitly forced dependency is primarily temporary, as in the not too distant future we will have a dependency on io.grpc:grpc-xds to support DirectPath.

Provides compiler validation of trying to concatenate two crc32c values

…etGetOptions...) (#1385)

* Introduce StorageArbitraries to contain common arbitraries which aren't globally registered by type.

Co-authored-by: BenWhitehead <BenWhitehead@users.noreply.github.com>

* Add Codec<DateTime, OffsetDateTime> for apiary conversion
* Add Utils#RFC_3339_DATE_TIME_FORMATTER
* Add Utils#millisOffsetDateTimeCodec to provide compatibility with existing Long centric apis
* Update BucketInfo and BucketInfo.Builder to prefer java.time types instead of Long
* Update ApiaryConversions for BucketInfo to use new java.time types

Update the following files to now be OffestDateTime instead of Long
* BlobInfo#customTime
* BlobInfo#deleteTime
* BlobInfo#updateTime
* BlobInfo#createTime
* BlobInfo#timeStorageClassUpdated
* BlobInfo#retentionExpirationTime
* BucketInfo#retentionEffectiveTime
* IamConfiguration#uniformBucketLevelAccessLockedTime

#1384)

* Add conversion for com.google.storage.v2.ServiceAccount
* Add property test to ensure round trip decode -> endode -> decode succeeds
* Add ServiceAccountArbitraryProvider

All Convertable types share a common set of properties that they need to provide. This new class will ensure in a central location they do.

First property: a round trip through the codec should provide a value equal to the initial input

Fixes an issue where upperbounds and flattend-pom were detecting lower versions via pubsub test deps.

### What's been added

Fields which map to apiary DateTime RFC 3339 fields will now also have
a preferred java.time.OffsetDateTime getter & setter.

Fields which map to apiary long to represent a duration in epoch millis
will no also have a preferred java.time.Duration getter & setter.

All new methods currently have the name of the corresponding java.time
type appended to the property name.

#### Refactor Long -> java.time.OffsetDateTime
* BlobInfo#createTime
* BlobInfo#customTime
* BlobInfo#deleteTime
* BlobInfo#retentionExpirationTime
* BlobInfo#timeStorageClassUpdated
* BlobInfo#updateTime
* Blob#createTime
* Blob#customTime
* Blob#deleteTime
* Blob#retentionExpirationTime
* Blob#timeStorageClassUpdated
* Blob#updateTime
* BucketInfo.CreatedBeforeDeleteRule
* BucketInfo.DeleteRule
* BucketInfo.IsLiveDeleteRule
* BucketInfo.NumNewerVersionsDeleteRule
* BucketInfo.RawDeleteRule
* BucketInfo#createTime
* BucketInfo#updateTime
* Bucket#createTime
* Bucket#updateTime
* Bucket#retentionEffectiveTime
* HmacKey.HmacKeyMetadata#createTime
* HmacKey.HmacKeyMetadata#updateTime

#### Refactor Long -> java.time.Duration
* BucketInfo#retentionPeriod
* Bucket#retentionPeriod

### Deprecation of existing methods
All existing "Long" based methods have been marked as `@Deprecated`
and had their javadocs updated to point to the new preferred
corresponding java.time method.

When the existing "Long" based methods are removed, the corresponding
java.time method will replace it at which point the java.time suffix
methods will be deprecated for later cleanup.

All existing "Long" based methods take care of performing the necessary
conversion to the respective java.time type and will continue to work
until their removal.

##### Why append the java.time type as a suffix?
1. Getters can't be overloaded, so we already need something to disambiguate
   which type should be returned.
2. For setters: we theoretically could overload since we have different
   method arguments, however, if a customer is already passing `null` as the
   method argument we've now created an ambiguity as to which method to
   invoke. By suffixing the setters as well, we avoid creating this possible
   ambiguity.

* Add StorageFixture to manage the test scoped lifecycle of a storage instance
* Update ITStorageTest to use new fixture
* Update ITBlobReadChannelTest to use new fixture

…ecated methods (#1415)

* Add GrpcConversions#timestampCodec convert between java.time.OffsetDateTime and com.google.protobuf.Timestamp
* Add property test for timestampCodec
* Replace all usage of "Long" methods in GrpcConversions with their appropriate java.time alternatives

Add initial implementation of a GrpcBlobWriteChannel and GrpcBlobReadChannel

### GrpcStorageImpl
GrpcStorageImpl#reader and GrpcStorageImpl#writer have been defined to construct
the respective channel.

Other read/write methods on GrpcStorageImpl have been implemented in terms of
`#reader` and `#writer`, more appropriate specific implementations will be
implemented later on.

#### Wired up methods
* `GrpcStorageImpl#create(BlobInfo blobInfo, BlobTargetOption... options)`
* `GrpcStorageImpl#create(BlobInfo blobInfo, byte[] content, BlobTargetOption... options)`
* `GrpcStorageImpl#create(BlobInfo blobInfo, byte[] content, int offset, int length, BlobTargetOption... options)`
* `GrpcStorageImpl#create(BlobInfo blobInfo, InputStream content, BlobWriteOption... options)`
* `GrpcStorageImpl#createFrom(BlobInfo blobInfo, Path path, BlobWriteOption... options)`
* `GrpcStorageImpl#createFrom(BlobInfo blobInfo, Path path, int bufferSize, BlobWriteOption... options)`
* `GrpcStorageImpl#createFrom(BlobInfo blobInfo, InputStream content, int bufferSize, BlobWriteOption... options)`
* `GrpcStorageImpl#readAllBytes(String bucket, String blob, BlobSourceOption... options)`
* `GrpcStorageImpl#readAllBytes(BlobId blob, BlobSourceOption... options)`
* `GrpcStorageImpl#reader(String bucket, String blob, BlobSourceOption... options)`
* `GrpcStorageImpl#reader(BlobId blob, BlobSourceOption... options)`
* `GrpcStorageImpl#downloadTo(BlobId blob, Path path, BlobSourceOption... options)`
* `GrpcStorageImpl#downloadTo(BlobId blob, OutputStream outputStream, BlobSourceOption... options)`
* `GrpcStorageImpl#writer(BlobInfo blobInfo, BlobWriteOption... options)`

### New channel session data types
`BlobWriteChannel`s existing implementation provides an internal handle to access
the resulting `StorageObject` which is created as part of a resumable upload
session. Additionally, performing a resumable upload is a multi-leg set of
operations. The new `*ByteChannelSession` interfaces and implementations
model this multi-leg lifecycle.

A `*ByteChannelSession` allows for opening of the channel for either read or write
along with keeping a reference to the underlying object associated with the
operation (the initial object including in the first message on read, or the
final object returned upon completing a write operation).

### New channels approach
Channels are inherently buffered already, in order to facilitate more efficient
and appropriate use of additional buffering all channels have Unbuffered
and Buffered implementations, where the buffered channels sole responsibility
is narrowed to the intermediate buffering otherwise delegating all operations
to an underlying unbuffered channel.

Buffers which are allocated for use with a BufferedChannel are appropriately
reused, rather than thrown away and/or resized in an ad-hoc manner. This should
provide a more predictable behavior for memory use.

All Channels now properly synchronize their read/write/flush methods to ensure
proper ordering from multiple threads.

### move over DataGeneration classes from benchmarks
Add DataGenerator, DataChain and TmpFile to test classes from benchmarks.

Various tools to dynamically generate data sets either in memory or on disk.

* chore: add initial skeleton of Retrying for Grpc

* rename RetryAlgorithmManager -> HttpRetryAlgorithmManager
* Add GrpcRetryAlgorithmManager
* Update Retrying.run(GrpcStorageOptions, ...) to take a Decoder instead
  of a Function (this helps narrow things a bit to our decoders)
* Add GrpcStorageImpl#SyntaxDecoders
  * For our "Syntax" types - Blob, Bucket, etc - Decoding an instance is
    always of the form message -> info -> syntax. This new class gives
    us a decoder that can use simply for message -> syntax by binding
    to the enclosing storage instance.
  * update "Syntax" return points to use new decoders

* feat: BucketInfo conversion for gRPC

* address feedback

* fix retentionPolicy encode

* address feedback

* lint

* use correct setter

* include proto-google-common-protos

Co-authored-by: BenWhitehead <BenWhitehead@users.noreply.github.com>

New property test against arbitrarily generated data along with buffer size, and app read size.

Ensure the expected number of reads and their sizes are returned for a specific data size.

Ensure channel is closed after full read, and throws a ClosedChannelException when attempted to read after close

…StorageImpl (#1439)


Add TransformPageDecorator which provides a Page implementation which knows how to transform
the result from the delegate page by using a specific mapping function.

Co-authored-by: BenWhitehead <BenWhitehead@users.noreply.github.com>

…d contract (#1446)

According to the official Javadocs, there is a contract that must exist between
#equals(Object)[1] and #hashCode()[2] for every class.

Prior to this change the contract was not upheld for the model classes due to
multiple reasons:
1. Some classes did not define equals or hashCode instead relying on the default
   implementations from java.lang.Object
2. Some classes were using different sets of fields between the two methods
3. Some classes were not performing deep evaluation
4. Some classes were serializing themselves to the apiary models before calling
   equals on the serialized instances

All model classes have been updated to ensure:
1. They define an equals and hashCode
2. The contract is upheld for these methods

Additionally, all internal usage of the deprecated DeleteRule has been removed
from BlobInfo. Instead, LifecycleRule are used exclusively. Deprecated
getDeleteRule and setDeleteRule have had their implementations changed to
perform an immediate conversion to/from LifecycleRule as necessary.

ApiaryConversions has been updated to ignore DeleteRule when encoding & decoding.

A codec for converting between DeleteRule <-> LifecycleRule has been added to
BackwardCompatibilityUtils. Corresponding tests for conversions between the
various types of DeleteRules have been added in BackwardCompatibilityUtilsTest.

Utils.ifNonNull has been updated to test the return value of the mapping
function before consuming it. Data.<Boolean>nullOf(Boolean.class) returns `true`
which can lead to unexpected behavior if a null value is provided to it
indirectly.

New Codec#andThen(Codec) has been added to allow easily connecting two Codecs.

[1] https://docs.oracle.com/javase/8/docs/api/java/lang/Object.html#equals-java.lang.Object-
[2] https://docs.oracle.com/javase/8/docs/api/java/lang/Object.html#hashCode--

test(deps): bump junit-platform.version to 5.9.1

Also removed a dangling bucket deletion in ITNotificationTest which was resulting in an exception at cleanup time.

…ll etags are still rolling out (#1693)

A Retention Periods duration is expressed in seconds, not milliseconds. Update test to ensure
the retention period duration that is processed is interpreted in the correct units.

* test: enable grpc config for ITBlobReadChannelTest

* fix: update GrpcBlobReadChannel to set limit as a number of bytes rather than end_offset

* fix: update StorageV2ProtoUtils#validation to treat limit as number of bytes to read rather than end_offset

* test: mark com.google.cloud.storage.it.ITBlobReadChannelTest.testReadChannelFailUpdatedGeneration as json only

* fix: update GrpcBlobReadChannel to match automatic decompression behavior of json

* fix: update GrpcBlobReadChannel to match exception behavior for failed preconditions

Split and renamed testReadChannelFail into the three scenarios it was testing
1. testReadChannel_preconditionFailureResultsInIOException_metagenerationMatch_specified
2. testReadChannel_preconditionFailureResultsInIOException_generationMatch_specified
3. testReadChannel_preconditionFailureResultsInIOException_generationMatch_extractor

Also update BlobReadChannel to remove RetryHelper$RetryHelperException from its cause chain

Add new builder method to all setting attempt of direct path connection.

If enabled, the underlying gapic client will be configured to attempt connecting via direct path.

deps: add a direct dependency to io.grpc:grpc-googleapis

Rumour has it this dependency will be removed from gax, and we need to pull it up to our level to ensure it's there for storage direct path.

…rc32c value (#1696)

Storage#createFrom and Storage#writer allow specifying an explicit crc32c value which should be passed to GCS for validation. Update our uploads to allow these explicitly defined values to be passed in the `finish_write` message to GCS.

Update UnifiedOpts.Crc32cValue to represent its value as an int rather than the b64 string from apiary. Internally when the value is necessary for apiary it will be converted to the b64 string representation.

Map gRPC Status `DATA_LOSS` to HTTP 400 Bad Request to mirror the behavior of JSON when a checksum validation fails.

Add a new Hasher.constant instant which will function to specify a constant value for the cumulative value of a write.

Various plumbing tweaks to allow passing this value all the way down and also attempting to "nullSafeConcat"

Add new integration tests to try the negative case to ensue we are getting checksum failures when we pass mis-match values and content.

* test: Enable GRPC for ITKmsTest

* test: enable gRPC config for ITAccessTest

* fix: fixup ITAccessTest#testRetentionPolicyNoLock to pass for gRPC

Introduce TemporaryBucket to provide an AutoClosable bucket which will cleanup on close

* fix: fixup ITAccessTest#testUBLAWithPublicAccessPreventionOnBucket to pass for gRPC

Split test into two different tests to cover each of the scenarios:
1. changingPAPDoesNotAffectUBLA
2. changingUBLADoesNotAffectPAP

* chore(test): fixup ITAccessTest generic acl and default acl tests to only run for JSON

* bucketAcl_requesterPays_false
* bucketAcl_requesterPays_true
* testBlobAcl
* testBucketDefaultAcl

* fix: fixup ITAccessTest related to Bucket IAM Policy

* testBucketPolicyV1
* testBucketPolicyV1RequesterPays
* testBucketPolicyV3

Add IamPolicyPropertyTest and corresponding provider.

* chore(test): set ITAccessTest#testBucketWithBucketPolicyOnlyEnabled to json only

* chore(test): set ITAccessTest#testBucketWithUniformBucketLevelAccessEnabled to json only

* chore: fmt

* fix: fixup ITAccessTest#testEnableAndDisableBucketPolicyOnlyOnExistingBucket to work with grpc instance

* fix: fixup ITAccessTest#testEnableAndDisableUniformBucketLevelAccessOnExistingBucket to work with grpc instance

* fix: fixup ITAccessTest#testEnforcedPublicAccessPreventionOnBucket to work with grpc instance

* fix: fixup ITAccessTest#testUnspecifiedPublicAccessPreventionOnBucket to work with grpc instance

…alues ahead of merge to main (#1704)

* Update javadocs on all new @BetaApi methods to state they are preview apis and subject to breaking changes
* Update TransportCompatibility annotations to apply to more classes and methods
* Regenerate all serialVersionUID values to ensure no leaking version numbers compared with other classes

…#1719)

… http codes (#1713)

Move all mapping from GrpcStorageImpl and BackwardCompatibilityUtils to GrpcToHttpStatusCodeTranslation.

Define new GrpcToHttpStatusCodeTranslation.StatusCodeMapping to explicitly define a mapping between io.grpc.Status.Code and our http code

Update tests to be explicit and not depend on utility methods which can also be used by the code under test.

Add new tests to ensure expected status codes are returned when evaluating DefaultStorageRetryStrategy

…shCode to include match prefix and suffix (#1729)

Add new test class ITBucketLifecycleRulesTest to validate some bucket mutations related to lifecycle rules behave as expected.

Update GrpcConversions to detect an empty list and preserve null value if the value from the server is empty.

* Define 3 new content size scenarios to be validated
  1. data is small enough to fit within a single message of a single stream
  2. data requires multiple messages within a single stream
  3. data requires multiple messages within multiple streams
* Refactor checksum validation centric tests out of ITObjectTest into new ITObjectChecksumSupportTest
* Move ITOptionRegressionTest.Content to top level as ChecksummedTestContent and flush it out a bit more for cross usage by ITObjectChecksumSupportTest

* fix: fix GrpcBlobReadChannel#isOpen

Previously a subbed method was still being used, now it will check the lazy channel.

Removed duplicate memoization of channel factory

* chore: update assertion for possible RPO values via gRPC