LLM Supervisor

Automatically switches OpenClaw between cloud and local Ollama models when rate limits occur.