Add mlflow span attribute logging for Genie ask_question inputs and outputs

[edurdevic]

• Added mlflow span attribute logging to Genie.ask_question for improved observability: logs space_id, input.question, input.conversation_id as input attributes and output.query, output.description, output.conversation_id as output attributes (intentionally excludes output.result to avoid logging potentially large query results).
• Added two unit tests verifying span attributes are correctly set for both new conversations and continued conversations.

Test plan

• test_ask_question_mlflow_trace_logs_inputs_and_outputs — verifies input/output attributes on a new conversation with a query attachment
• test_ask_question_mlflow_trace_with_conversation_id — verifies input/output attributes when continuing an existing conversation with a text attachment
• All 74 existing tests in test_genie.py continue to pass

Made with Cursor