fix: increase LLM timeout 30s→60s, fix Gateway request format, add Ollama optimization guide

- Fixed Gateway: 'prompt' → 'message' field name
- Increased LLM provider timeout from 30s to 60s
- Added OLLAMA-OPTIMIZATION.md with performance tips
- DAARWIZZ now responds (slowly but works)

providers/llm_provider.py

@@ -25,7 +25,7 @@ class LLMProvider(Provider):
         base_url: str,
         model: str,
         api_key: Optional[str] = None,
-        timeout_s: int = 30,
+        timeout_s: int = 60,
         max_tokens: int = 1024,
         temperature: float = 0.2,
         provider_type: str = "openai",  # "openai" or "ollama"