* py : fix internlm2-hf convert to gguf * ggml-ci
* support InternLM2 inference * add add_space_prefix KV pair