v1.3.2 - automatic captions work again

Automatic captions in DubForge broke for everyone who installed speech recognition after 29 September. This release fixes that.

Fixed

• "Generate captions" failed with open() got an unexpected keyword
  argument 'metadata_errors'. PyAV 19, a library faster-whisper uses to read
  audio, dropped an argument that faster-whisper still passes. DubForge now
  decodes the audio with its own ffmpeg, so captions no longer depend on it,
  and new installs keep PyAV below 19.

Updating

• Installed with the setup: run DubStage-Setup-1.3.2.exe below over your
  existing installation, or use the update banner in the app — the app-side fix
  alone is enough for captions to work again.
• Running from source: update, then
  py -m pip install -r requirements-transcription.txt.

Full history in CHANGELOG.md.