docs: document the extend_models client flag (merge inline models onto the catalog)
docs: prompt caching (prompt_cache, default on), token-usage prompt variables, cache pricing catalog fields
docs: Added documentation on the new automatic model catalog