Add an OllamaProvider that routes LLM calls to localhost:11434 instead
of OpenRouter. Ollama's API is compatible with the OpenAI format we already use.
This would allow Argus to run fully offline with local models like Gemma,
Llama, or Qwen. Suggested approach: add --provider ollama flag to argus-up.sh
and a matching provider variant in agent.rs
Add an OllamaProvider that routes LLM calls to localhost:11434 instead
of OpenRouter. Ollama's API is compatible with the OpenAI format we already use.
This would allow Argus to run fully offline with local models like Gemma,
Llama, or Qwen. Suggested approach: add --provider ollama flag to argus-up.sh
and a matching provider variant in agent.rs