Switch diarization model in Emilia to pyannote community-1

[gonzalo-cordova-pou]

✨ Description

This PR updates Emilia diarization to use pyannote/speaker-diarization-community-1 and adds compatibility handling so the pipeline works across pyannote API variants.

Changes:

• preprocessors/Emilia/main.py
  • Switch diarization model from pyannote/speaker-diarization-3.1 to pyannote/speaker-diarization-community-1.
  • Support both auth argument styles:
    • token=... (pyannote 4.x)
    • use_auth_token=... fallback (older versions)
  • Normalize diarization output handling:
    • use out.speaker_diarization when present
    • fallback to legacy Annotation output
• preprocessors/Emilia/README.md
  • Update model references from 3.1 to community-1.
  • Update environment example from Python 3.9 to 3.10.

Rationale:

• pyannote reports community-1 as improved over legacy 3.1 on most published benchmarks (with dataset-specific variance).

🚧 Related Issues

Closes #486

👨‍💻 Changes Proposed

• Migrate Emilia diarization model ID to community-1
• Add cross-version pyannote auth/output compatibility handling
• Update README setup/model references

🧑‍🤝‍🧑 Who Can Review?

@yuantuo666
@HarryHe11

✅ Checklist

• Code has been reviewed
• Code complies with the project's code standards and best practices
• Code has passed all tests
• Code does not affect the normal use of existing features
• Code has been commented properly
• Documentation has been updated (if applicable)
• Demo/checkpoint has been attached (if applicable)

Test Notes

Manual smoke test executed with a real audio sample and HF token:

• pipeline loads community-1
• diarization runs successfully
• output conversion path works (speaker_diarization/legacy fallback)
• first segment produced correctly