Add support for tool calling for models returned by ollama local

Author: mythz

llms/main.py

@@ -1400,6 +1400,7 @@ async def load_models(self):
                 "id": k,
                 "name": v.replace(":", " "),
                 "modalities": {"input": ["text"], "output": ["text"]},
+                "tool_call": True,
                 "cost": {
                     "input": 0,
                     "output": 0,