Remove MOSS-TTS-Nano backend (not real-time on free CPU tier)

ui: label engines honestly (VITS real-time / MOSS slow HD preview)

moss_ws: one-pass whole-text generation (fix inter-clause silence)

moss: official text normalization (WeTextProcessing) — fix numbers/entities

moss: drop persistent serve, use per-call moss-gen (fix free-tier hang)

moss_ws: revert to per-clause (one-pass hung the demo; restore working RTF 1.72)

moss_ws: one-pass generation for whole text (remove inter-utterance gaps)

moss: rebuild moss-gen with local KV cache (fork 09308c9, ~1.43x CPU)

moss: persistent moss-gen serve process (kill per-clause gguf reload)

moss: default to zh-TW voice (Yuewen/zh_4) when no user ref — fixes runaway generation

moss: rebuild moss-gen from fork rev dd64994 (MOSS_DIR support)

moss: capture moss-gen stderr to surface the real failure

moss: persist embedding tables to MOSS_DIR for moss-gen (fix exit 1)

moss_ws: robust worker (always close socket, surface errors, cap frames)

moss: warm the pipeline at startup (preload gguf into page cache)

moss: copy versioned ggml .so (libggml*.so*) — fix moss-gen loader

moss: add python-multipart (required by /moss_ref UploadFile) — fixes startup crash

moss: scope ggml libs to subprocesses (fix llama-cpp libggml clash)

moss: lazy init + background warm (don't block startup)

Add MOSS-TTS-Nano 2nd backend (rs.cpp + codec.cpp, ORT-free) with voice clone

Super-squash branch 'main' using huggingface_hub