LLM text chunks are streamed as message_file when assistant message has files

Start with api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py and trace QueueLLMChunkEvent handling into message_cycle_manager.py. Compare that path with QueueMessageFileEvent handling and review the unit test added in related PR #31414. Done means text chunks consistently produce message events while file metadata remains message_file events.

Self Checks

• I have read the Contributing Guide and Language Policy.
• This is only for bug report, if you would like to ask a question, please head to Discussions.
• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report, otherwise it will be closed.
• 【中文用户 & Non English User】请使用英语提交，否则会被关闭 ：）
• Please do not modify this template :) and fill in all the required fields.

Dify version

main branch at b96ea94505e3abe00daba69faa8bba9bcf9deb0a.

Cloud or Self Hosted

Self Hosted (Source) / upstream main code path.

Steps to reproduce

1. Use a chat/completion flow that streams an assistant response and also produces an assistant-side message file for the same message. For example, a workflow/tool path that emits QueueMessageFileEvent before or during LLM token streaming.
2. The file event is handled by MessageCycleManager.message_file_to_stream_response(), which caches the message id in _message_has_file and correctly emits a message_file event for the file metadata.
3. When a later QueueLLMChunkEvent is processed in EasyUIBasedGenerateTaskPipeline._process_stream_response(), the code calls MessageCycleManager.get_message_event_type(message_id=...) and reuses that event type for LLM text chunks.
4. Since the same message already has an assistant file, get_message_event_type() returns StreamEvent.MESSAGE_FILE.
5. The LLM text chunk is therefore emitted through message_to_stream_response(..., event_type=StreamEvent.MESSAGE_FILE) instead of StreamEvent.MESSAGE.

Current code evidence on main:

• api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py still precomputes the event type for QueueLLMChunkEvent by calling get_message_event_type(message_id=self._message_id) and passes it to message_to_stream_response().
• api/core/app/task_pipeline/message_cycle_manager.py returns StreamEvent.MESSAGE_FILE from get_message_event_type() when the message id is present in _message_has_file or when an assistant MessageFile exists for that message.

Related previously closed PR with a small fix and unit test: https://github.com/langgenius/dify/pull/31414

✔️ Expected Behavior

QueueLLMChunkEvent text chunks should always be streamed as event: message, regardless of whether the same assistant message also has files.

File metadata should continue to be streamed separately as event: message_file via QueueMessageFileEvent.

❌ Actual Behavior

When the assistant message has a file, LLM text chunks can be streamed as event: message_file. The web client may treat those chunks as file events instead of normal message text, so streamed answer text can be dropped or not rendered correctly.