Why not support any OpenAI-compatible provider?

[yookibooki]

Request: Add a generic openai-compatible provider that users can configure in ~/.keen/configs.json with a base_url, api_key, and headers.
Expected config (example):

{
  "active_provider": "inference.com",
  "active_model": "glm-5.2",
  "providers": {
    "inference.com": {
      "base_url": "https://api.inference.com/v1",
      "api_key": "sk-...",
      "models": ["glm-5.2", "minimax-m3"]
    }
  }
}