Rename ipython to tool (#756)

See https://github.com/meta-llama/llama-models/pull/261 for the
corresponding PR in llama-models.

Once these PRs land, you need to work `main` from llama-models (vs. from
pypi)

llama_stack/apis/inference/inference.py

@@ -82,7 +82,7 @@ class SystemMessage(BaseModel):
 
 @json_schema_type
 class ToolResponseMessage(BaseModel):
-    role: Literal["ipython"] = "ipython"
+    role: Literal["tool"] = "tool"
     # it was nice to re-use the ToolResponse type, but having all messages
     # have a `content` type makes things nicer too
     call_id: str

llama_stack/providers/remote/inference/nvidia/openai_utils.py

@@ -144,7 +144,7 @@ def _convert_message(message: Message | Dict) -> OpenAIChatCompletionMessage:
             message = UserMessage(**message)
         elif message["role"] == "assistant":
             message = CompletionMessage(**message)
-        elif message["role"] == "ipython":
+        elif message["role"] == "tool":
             message = ToolResponseMessage(**message)
         elif message["role"] == "system":
             message = SystemMessage(**message)

tests/client-sdk/agents/test_agents.py

@@ -40,10 +40,10 @@ class TestClientTool(ClientTool):
             response_str = f"Error when running tool: {e}"
 
         message = ToolResponseMessage(
+            role="tool",
             call_id=tool_call.call_id,
             tool_name=tool_call.tool_name,
             content=response_str,
-            role="ipython",
         )
         return [message]