Open-dubbing

Added
- Support for VAD filter when using fasterwhisper
- Implemented retry when using Apertium translation API

Fixed
- Better ffmpeg error login and robust execution

Added
- Ability to add original and dubbed subtitles as stream in the output video

Fixed
- Reduced the number of warnings from 3rd party libraries
- Switch to ffmpeg to adjust the audio speed since pydub did not work in some cases

Fixed
- Standardize the naming of some cli parameters

Added
- Retry if Edge TTS fails to provide the synthesis. Happens sometimes
- Improved 'update' command which updates now utterances file and checks files needed

Fixed
- Remove empty blocks of dubbed audios that do not contain text
- Speed calculation: when it's the last block, not to increase the speed if is not needed

Added
- Support for manually postediting the automatically dubbing (--update)
- Support for Whisper large-v2 model (better than v3 for some languages)

Fixed
- Do not need to merge back audios that have not been dubbed
- If a file merge file fails, do not fail the whole batch