Add VibeVoice-ASR backend

Third ASR backend option alongside faster-whisper and qwen3. Builds
microsoft/VibeVoice from source (pinned to a specific commit, since it's
custom modeling code not in transformers' Auto* registry) into its own venv,
inheriting the base image's torch/CUDA. Comes with built-in speaker
diarization (VibeVoiceASRProcessor.post_process_transcription returns
per-segment speaker ids directly, no separate pyannote pass needed).

Verified end-to-end against a real recording: 200 OK, correct Korean
transcription, speaker labels populated.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

app/supervisord.conf

@@ -26,6 +26,17 @@ stderr_logfile=/dev/fd/2
 stderr_logfile_maxbytes=0
 priority=20
 
+[program:vibevoice]
+command=/opt/venvs/vibevoice/bin/python /app/asr/workers/vibevoice_worker.py --host 0.0.0.0 --port 8006
+directory=/app
+autostart=true
+autorestart=true
+stdout_logfile=/dev/fd/1
+stdout_logfile_maxbytes=0
+stderr_logfile=/dev/fd/2
+stderr_logfile_maxbytes=0
+priority=25
+
 [program:xtts]
 command=/opt/venvs/xtts/bin/python /app/tts/workers/xtts_worker.py --host 0.0.0.0 --port 8005
 directory=/app