Update llama.cpp related code to use openai format

[ptelang]

This PR modifies the llama.cpp related implementation to use openai format request/response.

It assumes that Continue plugin is configured as follows.

"models": [
...
  {
      "title": "Llama CPP",
      "provider": "openai",
      "model": "qwen2.5-coder-1.5b-instruct-q5_k_m",
      "apiBase": "http://localhost:8989/llamacpp"
  },
...
],
...
"tabAutocompleteModel": {
    "title": "Llama CPP",
    "provider": "openai",
    "model": "qwen2.5-coder-1.5b-instruct-q5_k_m",
    "apiBase": "http://localhost:8989/llamacpp",
},
...

[jhrozek]

The code looks good and works fine. Thank you!