docs(site): add SenseVoice audio.cpp deployment

[LauraGPT]

Summary

• expand the audio.cpp deployment contract from Fun-ASR-Nano-only to stable Fun-ASR-Nano plus the SenseVoice candidate runtime
• pin the SenseVoice path to audio.cpp commit b748ca509adc16c15aff44f76456fd47b257c933 and the official schema-v1 GGUF revision
• document offline CLI, buffered streaming CLI/SSE, OpenAI-compatible API, verification evidence, limitations, and the GGUF SHA-256
• keep the candidate status explicit until audio.cpp PR a bug in funasr/models/pooling/statistic_pooling.py:85 #219 is merged

Why

The existing product page said audio.cpp supported only Fun-ASR-Nano and did not support streaming. That became incomplete after the SenseVoice schema-v1 implementation passed all six upstream CI jobs. Users need one accurate deployment contract that distinguishes the merged Nano path from the pinned SenseVoice candidate instead of following a floating branch.

Validation

• python3 -m pytest web-pages/product-site/tests -q (68 passed)
• generated 25 product pages with web-pages/product-site/build.py
• web-pages/product-site/validate.py validated 100 HTML pages
• checked both generated audio.cpp pages for bilingual titles, exact candidate commit, 6/6 CI status, install/streaming commands, pinned HF revision, GGUF SHA-256, and evidence links

Evidence

• audio.cpp candidate: Add SenseVoice offline and streaming ASR (schema v1) 0xShug0/audio.cpp#219
• pinned GGUF: https://huggingface.co/FunAudioLLM/SenseVoiceSmall-GGUF-audiocpp/tree/5c3fcfe748a8714216bc135476d5863084fddb72
• GGUF SHA-256: 4dedf169f625437fb336f2959674f399819729a765e184128c0e25a6e16ff0ec