feat(runtime): add GGUF HTTP transcription wrapper

[LauraGPT]

Summary

Adds a lightweight OpenAI-compatible HTTP wrapper for the existing llama.cpp / GGUF command-line binaries.

• new runtime/llama.cpp/server/funasr_gguf_server.py standard-library server
• accepts POST /v1/audio/transcriptions multipart uploads and returns {\"text\": ...}
• forwards requests to the configured llama-funasr-* binary with -m, -a, optional --vad, optional --backend, and extra forwarded args
• documents local startup and curl usage in runtime/llama.cpp/README.md

Fixes #3298.

Validation

• python3 -m pytest -q runtime/llama.cpp/tests/test_gguf_server.py runtime/llama.cpp/tests/test_release_workflow.py runtime/llama.cpp/tests/test_backend_flags.py -q
• python3 -m py_compile runtime/llama.cpp/server/funasr_gguf_server.py
• git diff --check

[gemini-code-assist]

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.