Speechkitty

* New script ``sample/transcribe_with_diarization.py`` performs transcription with diarization using [whisperX](https://github.com/AlekseiPrishchepo/whisperX-REST) backend,
* Speaker tags for _whisperX_ backend added.
* Credentials now can be provided for ``Transcribe()`` two ways: in parameters or in ``.env``.

* Added [whisperX-REST](https://github.com/AlekseiPrishchepo/whisperX-REST) (RESTful API for whisperX) for the backend.
* Added ``sample/transcribe_split_channels.py`` to transcribe split channels (since whisperX only transcribes one audio channel).
* Credentials moved to ``.env``.
* Other improvements.

Class ``Transcriber``:
- new method ``transcribe_file()`` encompasses transcribing steps from uploading file to fetching json;
- new method ``set_raise_exceptions()`` controls whether to continue if an error occurred;
- 100% tests coverage (still not yet tested on audio recordings with different characteristics).

Other:
- switched from ``setup.py`` to ``pyproject.toml``.

Class ``Directory``:

- ``get_wavs()`` has been renamed to ``get_records()`` as well as its default ``regexp_include`` parameter changed to search for common audio files extensions.

Class ``Transcriber``:

- ``wav_to_ogg()`` has been renamed to ``to_ogg()`` and accepts all the audio formats that ffmpeg is able to convert to **ogg opus**.

Other:

- ``transcribe_directory.py`` first checks ``credentials.ini`` in its directory and then looks for environment variables.

Class ``Transcriber``:
- new method ``transcribe_file()`` encompasses transcribing steps from uploading file to fetching json;
- new method ``set_raise_exceptions()`` controls whether to continue if an error occurred;
- 100% tests coverage (still not yet tested on audio recordings with different characteristics).