Add message_id to LLM events to differentiate stream chunks

[devin-ai-integration]

Add message_id to LLM events to differentiate stream chunks

Summary

This PR fixes issue #3845 by adding a unique message_id field to all LLM events. Previously, when an agent made multiple LLM calls within the same task, all LLMStreamChunkEvent chunks shared the same task_id and agent_id, making it impossible to differentiate which chunks belonged to which LLM call.

Changes:

• Added message_id: str | None = None field to LLMEventBase (maintains backward compatibility)
• Generate unique message_id (UUID) at the start of each LLM.call()
• Thread message_id through all code paths (streaming, non-streaming, tool calls, error handling)
• All LLM events now include the same message_id for events from the same call
• Added comprehensive test suite to verify message_id uniqueness and consistency

Review & Testing Checklist for Human

⚠️ Risk Level: YELLOW - Local testing was blocked by environment issues

• Run tests locally: Tests were written but not run locally due to uv.lock corruption. Please run uv run pytest lib/crewai/tests/test_llm_message_id.py -vv to verify they pass
• Verify streaming behavior: Make actual streaming LLM calls from the same agent/task and confirm that events from different calls have different message_id values
• Check event consistency: Verify that all events (Started, StreamChunk, Completed) from a single LLM call share the same message_id
• Backward compatibility: Ensure existing event listeners that don't expect message_id still work correctly
• Code review: Scan through the diff to ensure all function signatures were updated consistently and no call sites were missed

Test Plan

1. Create an agent and task
2. Make multiple LLM calls with streaming enabled
3. Register event listeners for LLMCallStartedEvent and LLMStreamChunkEvent
4. Verify that each call gets a unique message_id
5. Verify that all chunks within one call share the same message_id

Notes

• The message_id field is optional (defaults to None) for backward compatibility
• All existing tests should continue to pass since the field is optional
• Link to Devin run: https://app.devin.ai/sessions/4df91e38125b455ca8fd1ea091a8245b
• Requested by: João (joao@crewai.com)

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring