Add OpenAI provider support

[ahmadhozien]

Closes #206

What changed

• added provider-aware config helpers for Ollama and OpenAI
• added OpenAI Responses API support to the LLM adapter
• updated startup and cron model selection to pass the provider through
• documented the new OpenAI config flow in the docs and example config

Validation

• venv\\Scripts\\python.exe -m py_compile src\\config.py src\\llm_provider.py src\\main.py src\\cron.py

[kamusis]

Hey @ahmadhozien, thanks for the contribution! One concern with the current implementation: generate_text calls POST /responses (OpenAI's Responses API), which is OpenAI-specific and not part of the standard OpenAI-compatible API spec.

Most providers that advertise OpenAI compatibility — such as OpenRouter, MiniMax, Groq, Azure OpenAI, and others — only implement POST /v1/chat/completions, not /responses. This means setting a custom openai_base_url to any of those providers will still fail.

It would be much better to use POST /chat/completions with the standard payload:

{
  "model": "<model>",
  "messages": [{"role": "user", "content": "<prompt>"}]
}

This is universally supported across all OpenAI-compatible APIs and would make the openai_base_url option genuinely useful for third-party providers. Would you be open to switching to that endpoint?