A listener for the aggregate state changes of the services that are under management. Users
that need to listen to more fine-grained events (such as when each particular
service starts, or terminates), should attach service listeners to each individual service.
Called when the service initially becomes healthy.
This will be called at most once after all the services have entered the
running state. If any services fail during start up or
fail/terminate before all other
services have started running then this method will not be called.