feat: add LuxTTS as second TTS engine with multi-engine support

Introduce LuxTTS (ZipVoice) alongside Qwen TTS, enabling users to choose
between engines at generation time. LuxTTS offers fast, English-focused
voice cloning at 48kHz with ~1GB VRAM.

Backend:
- Add LuxTTSBackend with encode_prompt/generate_speech integration
- Multi-engine registry (get_tts_backend_for_engine) replacing singleton
- Engine-prefixed voice prompt cache keys to avoid collisions
- Engine field on GenerationRequest (default 'qwen' for backward compat)
- Engine dispatch in /generate and /generate/stream endpoints
- LuxTTS in model status, download, and delete maps

Frontend:
- TTS Engine selector dropdown in GenerationForm (Qwen TTS / LuxTTS)
- Conditionally hide Model Size and Delivery Instructions for LuxTTS
- Engine field added to TypeScript types and Zod schema
- LuxTTS section in Model Management page

backend/requirements.txt

@@ -9,11 +9,19 @@ alembic>=1.13.0
 
 # ML models
 torch>=2.1.0
-transformers>=4.36.0
+transformers>=4.36.0,<=4.57.6
 accelerate>=0.26.0
 huggingface_hub>=0.20.0
 qwen-tts>=0.0.5
 
+# LuxTTS (voice cloning engine)
+Zipvoice @ git+https://github.com/ysharma3501/LuxTTS.git
+onnxruntime>=1.16.0
+piper-phonemize>=1.1.0
+lhotse>=1.20.0
+pydub>=0.25.0
+inflect>=7.0.0
+
 # Audio processing
 librosa>=0.10.0
 soundfile>=0.12.0