feat(responses)!: add in_progress, failed, content part events (#3765)

## Summary
- add schema + runtime support for response.in_progress /
response.failed / response.incomplete
- stream content parts with proper indexes and reasoning slots
- align tests + docs with the richer event payloads

## Testing
- uv run pytest
tests/unit/providers/agents/meta_reference/test_openai_responses.py::test_create_openai_response_with_string_input
- uv run pytest
tests/unit/providers/agents/meta_reference/test_response_conversion_utils.py

llama_stack/apis/agents/openai_responses.py

@@ -438,7 +438,7 @@ class OpenAIDeleteResponseObject(BaseModel):
 class OpenAIResponseObjectStreamResponseCreated(BaseModel):
     """Streaming event indicating a new response has been created.
 
-    :param response: The newly created response object
+    :param response: The response object that was created
     :param type: Event type identifier, always "response.created"
     """
 
@@ -446,11 +446,25 @@ class OpenAIResponseObjectStreamResponseCreated(BaseModel):
     type: Literal["response.created"] = "response.created"
 
 
+@json_schema_type
+class OpenAIResponseObjectStreamResponseInProgress(BaseModel):
+    """Streaming event indicating the response remains in progress.
+
+    :param response: Current response state while in progress
+    :param sequence_number: Sequential number for ordering streaming events
+    :param type: Event type identifier, always "response.in_progress"
+    """
+
+    response: OpenAIResponseObject
+    sequence_number: int
+    type: Literal["response.in_progress"] = "response.in_progress"
+
+
 @json_schema_type
 class OpenAIResponseObjectStreamResponseCompleted(BaseModel):
     """Streaming event indicating a response has been completed.
 
-    :param response: The completed response object
+    :param response: Completed response object
     :param type: Event type identifier, always "response.completed"
     """
 
@@ -458,6 +472,34 @@ class OpenAIResponseObjectStreamResponseCompleted(BaseModel):
     type: Literal["response.completed"] = "response.completed"
 
 
+@json_schema_type
+class OpenAIResponseObjectStreamResponseIncomplete(BaseModel):
+    """Streaming event emitted when a response ends in an incomplete state.
+
+    :param response: Response object describing the incomplete state
+    :param sequence_number: Sequential number for ordering streaming events
+    :param type: Event type identifier, always "response.incomplete"
+    """
+
+    response: OpenAIResponseObject
+    sequence_number: int
+    type: Literal["response.incomplete"] = "response.incomplete"
+
+
+@json_schema_type
+class OpenAIResponseObjectStreamResponseFailed(BaseModel):
+    """Streaming event emitted when a response fails.
+
+    :param response: Response object describing the failure
+    :param sequence_number: Sequential number for ordering streaming events
+    :param type: Event type identifier, always "response.failed"
+    """
+
+    response: OpenAIResponseObject
+    sequence_number: int
+    type: Literal["response.failed"] = "response.failed"
+
+
 @json_schema_type
 class OpenAIResponseObjectStreamResponseOutputItemAdded(BaseModel):
     """Streaming event for when a new output item is added to the response.
@@ -688,19 +730,46 @@ class OpenAIResponseObjectStreamResponseMcpCallCompleted(BaseModel):
 
 @json_schema_type
 class OpenAIResponseContentPartOutputText(BaseModel):
+    """Text content within a streamed response part.
+
+    :param type: Content part type identifier, always "output_text"
+    :param text: Text emitted for this content part
+    :param annotations: Structured annotations associated with the text
+    :param logprobs: (Optional) Token log probability details
+    """
+
     type: Literal["output_text"] = "output_text"
     text: str
-    # TODO: add annotations, logprobs, etc.
+    annotations: list[OpenAIResponseAnnotations] = Field(default_factory=list)
+    logprobs: list[dict[str, Any]] | None = None
 
 
 @json_schema_type
 class OpenAIResponseContentPartRefusal(BaseModel):
+    """Refusal content within a streamed response part.
+
+    :param type: Content part type identifier, always "refusal"
+    :param refusal: Refusal text supplied by the model
+    """
+
     type: Literal["refusal"] = "refusal"
     refusal: str
 
 
+@json_schema_type
+class OpenAIResponseContentPartReasoningText(BaseModel):
+    """Reasoning text emitted as part of a streamed response.
+
+    :param type: Content part type identifier, always "reasoning_text"
+    :param text: Reasoning text supplied by the model
+    """
+
+    type: Literal["reasoning_text"] = "reasoning_text"
+    text: str
+
+
 OpenAIResponseContentPart = Annotated[
-    OpenAIResponseContentPartOutputText | OpenAIResponseContentPartRefusal,
+    OpenAIResponseContentPartOutputText | OpenAIResponseContentPartRefusal | OpenAIResponseContentPartReasoningText,
     Field(discriminator="type"),
 ]
 register_schema(OpenAIResponseContentPart, name="OpenAIResponseContentPart")
@@ -710,15 +779,19 @@ register_schema(OpenAIResponseContentPart, name="OpenAIResponseContentPart")
 class OpenAIResponseObjectStreamResponseContentPartAdded(BaseModel):
     """Streaming event for when a new content part is added to a response item.
 
+    :param content_index: Index position of the part within the content array
     :param response_id: Unique identifier of the response containing this content
     :param item_id: Unique identifier of the output item containing this content part
+    :param output_index: Index position of the output item in the response
     :param part: The content part that was added
     :param sequence_number: Sequential number for ordering streaming events
     :param type: Event type identifier, always "response.content_part.added"
     """
 
+    content_index: int
     response_id: str
     item_id: str
+    output_index: int
     part: OpenAIResponseContentPart
     sequence_number: int
     type: Literal["response.content_part.added"] = "response.content_part.added"
@@ -728,15 +801,19 @@ class OpenAIResponseObjectStreamResponseContentPartAdded(BaseModel):
 class OpenAIResponseObjectStreamResponseContentPartDone(BaseModel):
     """Streaming event for when a content part is completed.
 
+    :param content_index: Index position of the part within the content array
     :param response_id: Unique identifier of the response containing this content
     :param item_id: Unique identifier of the output item containing this content part
+    :param output_index: Index position of the output item in the response
     :param part: The completed content part
     :param sequence_number: Sequential number for ordering streaming events
     :param type: Event type identifier, always "response.content_part.done"
     """
 
+    content_index: int
     response_id: str
     item_id: str
+    output_index: int
     part: OpenAIResponseContentPart
     sequence_number: int
     type: Literal["response.content_part.done"] = "response.content_part.done"
@@ -744,6 +821,7 @@ class OpenAIResponseObjectStreamResponseContentPartDone(BaseModel):
 
 OpenAIResponseObjectStream = Annotated[
     OpenAIResponseObjectStreamResponseCreated
+    | OpenAIResponseObjectStreamResponseInProgress
     | OpenAIResponseObjectStreamResponseOutputItemAdded
     | OpenAIResponseObjectStreamResponseOutputItemDone
     | OpenAIResponseObjectStreamResponseOutputTextDelta
@@ -763,6 +841,8 @@ OpenAIResponseObjectStream = Annotated[
     | OpenAIResponseObjectStreamResponseMcpCallCompleted
     | OpenAIResponseObjectStreamResponseContentPartAdded
     | OpenAIResponseObjectStreamResponseContentPartDone
+    | OpenAIResponseObjectStreamResponseIncomplete
+    | OpenAIResponseObjectStreamResponseFailed
     | OpenAIResponseObjectStreamResponseCompleted,
     Field(discriminator="type"),
 ]