Faster AI chat responses with caching and provider health checks

Improved AI chat responsiveness with statistics caching, agent configuration caching, RAG search caching, provider health checks, and async token logging, reducing typical response times significantly.