Queries regarding "User simulator based evaluation" approach in adk

[AmitKumarShorthillsAI]

I was doing the evaluation of my agents and had something to clarify because i cannot find detailed explanation in the documentation:

• If my agent is deployed on vertex, then how can i communicate with it for user-simulation based evaluation?
• I only see hallucination and security based metrics, how can i define my own metrics, where will the logic of those metric reside?
• Does the current version allows us to run multiple test cases in parallel? How could i run multiple test cases in parallel? Is there any concurrent tests run like method in user-simulation based evaluation?

[AmitKumarShorthillsAI]

Also, can you suggest how can i log particular things like User message, Agent response, tool/function calls etc when i run user simulation through "adk eval" cli tool?