As a StudioML user, I want to be able to use persistent workload queues, so that queue server restart will not disrupt my experiment flow.