auto start in large deployments

[coog83]

Autostart in large environments are challenging such that don't overload a satellite trying to start too many nodes at ones across N labs. In the future can you implement a message bus like Kafka or rabbitmq that can handle this where the master / gui puts the start request on the bus then each satellite will grab start requests off the bus when it is ready to start another where if it can start 5, the 6th sits in on the bus until one of the 5 is fully started. This would organize the chaos of the startups across multiple labs first thing in the morning.