Use service group for service capabilities

[Joshua Harlow]

Instead of periodically sending out compute node 'capabilities' MQ messages to schedulers it might make sense to use the new service group concept and store information about the service there, since the service group concept is already used to check if a service is 'available' by the scheduler, why not just store information along with said 'liveness' about what the service can offer.

At least in the case of zookeeper 'nodes' can store information, and since these nodes are already being used for liveness (by the existence of a node) it seems to make sense to store other data there (or at least have the potential to?)