Introduce a new `PyPDF2.generic.Fit` class which captures the type and parameter for how a page should be fit in the viewer (e.g. when clicking on a PDF-internal link).

The class has one method for each fit type which allows users to discover the different types via their IDE, e.g. `Fit.xyz(left=123, top=456, zoom=2)`.

## Breaking Change

This **introduces a breaking change** that needs a major version bump. Affected are the following methods:

* `PdfMerger.add_outline_item` used `fit: FitType` and `*args: ZoomArgType` parameters
* `PdfWriter.add_outline_item` used `fit: FitType` and `*args: ZoomArgType` parameters
* `AnnotationBuilder.link`  used `fit: FitType = "/Fit"` and `fit_args: Tuple[ZoomArgType, ...] = tuple()` instead.

Instead of having two arguments, we now have only one. To that argument, an object of the new `Fit` class must be passed.

## Why *args is problematic

Using the `*args` pattern is problematic for two reasons:

* **User-code readability**:
  * People cannot use the more expressive keyword-argument syntax in their code for the non-fit parameters
  * People may or may not use the keyword-parameter for the fit parameters
* **Library extensions**: PyPDF2 cannot easily add new parameter to those functions; #1371 is the latest PR that stumbled over this issue
* **Two parameter for one thing**: In general I don't like if we have two parameters for one topic

Co-authored-by: pubpub-zz <4083478+pubpub-zz@users.noreply.github.com>

The method `.clone(pdf_dest,[force_duplicate])` clones the objects and all referenced objects.

If an object is already cloned, the already cloned object is returned (unless force_duplicate is set)
mainly for internal use but can be used on a page
for pageObject/DictionnaryObject/[Encoded/Decoded/Content]Stream an extra parameter ignore_fields list that provide the list of fields that should not be cloned.

When available, the pointer to an object is available in `indirect_obj` attribute.

New API for add_page/insert_page that :

* returns the cloned page object
* ignore_fields can be provided as a parameter.

## Others

* file is closed at the end of PdfWriter.write when a filename is provided
* Breaking Change: `add_outline_item` now has a parameter before which is not the last parameter

## Update
* The public API of PdfMerger has been added to PdfWriter (ready to make PdfMerger an alias of it)
* Process properly Outline merging
* Process properly Named destinated

Deals with #1194, #1322, #471, #1337

Closes #1494

In addition, the following changes were made:

* DEV: Add in-project virtual envs to .gitignore

Many developers (like myself) like to use virtual environments included within
the current project. These virtual environment are local development constructs
and should not be checked into source control.

This commit adds two common virtual environment directory names to the
.gitignore to avoid accidental commits from future developers.

* DEV: Include `pillow` in `requirements/dev.in`

The current contribution instructions in `docs/dev/intro.md` direct new code
contributors to install the `dev` requirements. After following that
instruction, the minimal test suite fails.

This commit adds `pillow` to  `requirements/dev.in` so that the minimal test
suite can pass on the first try so that new code contributors can start
implementing improvements with confidence.

* STY: Use official `IO` type for file streams

The Python standard library provides the `IO` type for file streams. (Source:
https://docs.python.org/3/library/typing.html#typing.IO)

This commit replaces the complex Union type of the `IO` implementations with the
official `IO` type. This will improve the accuracy of type checking in users'
IDEs.

* STY: Use standard `IO` type hint for writers

The CI system flagged some additional conflicts with the `IO` type in the writer
classes.

This commit changes the writer classes to use the standard `IO` type instead of
the union of IO implementations.

Deprecate features, but keep helpful exceptions. That means that the names still need to be there.

The deprecated names will stay until 4.0.0 to help the community transition to the new names.

See https://myst-parser.readthedocs.io/en/latest/configuration.html

Closes #1492

BREAKING CHANGES:
-  Deprecate features with PyPDF2==3.0.0 (#1489)
-  Refactor Fit / Zoom parameters (#1437)

New Features (ENH):
-  Add Cloning  (#1371)
-  Allow int for indirect_reference in PdfWriter.get_object (#1490)

Documentation (DOC):
-  How to read PDFs from S3 (#1509)
-  Make MyST parse all links as simple hyperlinks (#1506)
-  Changed 'latest' for 'stable' generated docs (#1495)
-  Adjust deprecation procedure (#1487)

Maintenance (MAINT):
-  Use typing.IO for file streams (#1498)

[Full Changelog](2.12.1...3.0.0)