implement code build

add detect function to CodeBuild ci provider

Recently testing the CLI with the new upload endpoint I had the following error (when trying to upload the CLI itself)
```
TypeError: keys must be str, int, float, bool or None, not PosixPath
Error: Process completed with exit code 1.
```

After some digging I found the keys in question
```
Finding the PosixPath
Key languages/languages.c is PosixPath. Origin: .path_fixes.value[languages/languages.c]
Key tests/data/files_to_fix_examples/sample.cpp is PosixPath. Origin: .path_fixes.value[tests/data/files_to_fix_examples/sample.cpp]
Key tests/data/files_to_fix_examples/sample.go is PosixPath. Origin: .path_fixes.value[tests/data/files_to_fix_examples/sample.go]
Key tests/data/files_to_fix_examples/sample.kt is PosixPath. Origin: .path_fixes.value[tests/data/files_to_fix_examples/sample.kt]
Key tests/data/files_to_fix_examples/sample.php is PosixPath. Origin: .path_fixes.value[tests/data/files_to_fix_examples/sample.php]
Finding the PosixPath ================ END
```

They come from the path_fixes, which are of class Path, but need to be strings.
To solve this we simply cast them to string while building the payload.

The reason this was not picked up in the unit tests before is that we were already passing strings, not Path as it should be.

Recently when trying out the CLI with the new upload endpoint I tried to use the `--exclude` option but got the following error:
```
File "/Users/giovannimguidini/Projects/GitHub/codecov-cli/codecov_cli/services/upload/coverage_file_finder.py", line 179, in find_coverage_files
    default_folders_to_ignore + self.folders_to_ignore,
TypeError: can only concatenate list (not "tuple") to list
```

Investigating it deeper we try to merge 2 lists, with one of them being actually a tuple indeed:
```
default_folders_to_ignore has type <class 'list'>
self.folders_to_ignore has type <class 'tuple'>
```

Click documentation doesn't specify if the default cliss for options with `multiple=True` is `tuple` or `list`, nor did I find a click type for it. So instead I'm just casting the argument passed by click to be sure. I changed the default so it would still work if we don't use those options.

We have setup a workflow for releasing codecov-cli. Currently it runs on every push to master. If you look at the releases page there are MANY releases being created as drafts. That polutes the visibility of true releases. Furthermore, the only asset we have being built works only on Linux distros, and we'd like to provide something for macos and windows users as well.

We also have redundant CI jobs for PRs and master on push.

To address all those concerns, first we are updating the CI workflows. `push_flow.yml` will run the lint-build-test steps for every push we have. This also uploades test coverage to codecov. `release_flow.yml` takes care of building new releases. It only runs when a new TAG is pushed, starting with `v` (for versions).

To help us use the new release flow a new command is begin added to Makefile, the `tag.release` command. It should be used as `make tag.release version=v0.1.0`, for example. It takes care of tagging a new release and uploading it to GitHub, hopefully triggering the release_flow as well.

The changes for `release_flow.yml` were inspired by [this tutorial](https://data-dive.com/multi-os-deployment-in-cloud-using-pyinstaller-and-github-actions/)

…illy to draft (#105)

* Include purge release step
* Add changelog for release
* Fix issue when creating final release file we had. That was probably because we were using `github.ref` which contains the entire name (e.g. `refs/tags/release`) not simply `release`. Changed to `github.ref_name` to be what we expect.

Rename `build-assets` to `buildassets` for consistency and fix syntax error