feat: integrate training UI, BEATs model, and clean up legacy code

- Remove legacy distance-mode scanning (build_profile, _similarity, etc.)
  and hand-crafted intensity features — pipeline is now embedding-only
- Integrate Microsoft BEATs as embedding option alongside wav2vec2/HuBERT
- Add TrainDialog with positive class selector, model picker, video dir
  fallback, and live training stats
- Add TrainWorker QThread with cancel support and proper lifecycle cleanup
- Add source_path column to DB for robust source video tracking
- Add get_export_folders/get_training_data/get_training_stats to DB
- Wire source_path in all export DB writes (_on_clip_done, _on_auto_clip_done)
- Cancel scan/train workers in closeEvent to prevent use-after-free crashes
- Add setup_env.sh supporting both conda and python venv (CUDA 12.8)
- Update requirements.txt with all actual dependencies
- Update 8cut_train.py with --positive flag for new DB-driven training

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

requirements.txt

@@ -1,4 +1,25 @@
+# Core GUI
 PyQt6>=6.4
 python-mpv>=1.0
-pytest>=7.0
+
+# Audio & ML
+librosa>=0.10
+numpy>=1.24
+scikit-learn>=1.3
+joblib>=1.3
+soundfile>=0.12
+
+# Deep learning (torch installed separately for CUDA support)
+# torch and torchaudio are installed via --index-url in setup_env.sh
+torchaudio>=2.0
+
+# Object detection
 ultralytics>=8.0
+
+# Server API
+fastapi>=0.100
+pydantic>=2.0
+uvicorn>=0.23
+
+# Dev
+pytest>=7.0