Description
This is important because agent might crash after the executor issued 'wait'. The executor will like to issue another 'wait' to monitor its nested container. If the nested container terminates in between, currently, we don't give back the correct exit status of the nested container.
This is not nice because the behavior of the executor will depend on if the agent is up or not. We should expect executor to get exit status for an already terminated nested container and we should return the correct exit status.