Feature: Expose more service/replica level metrics

[heapdavid]

Hi,

We're currently monitoring docker swarm with prometheus and we're looking for a way to check for services not being fulfilled/replicas not running.

We can see metrics such as swarm_manager_services_total and failed/rejected containers but would ideally love something like swarm_manager_replicas_total with e.g. state=running and state=desired labels for comparison/alerting.

This could be a summation of the data presented under the REPLICAS column when one runs docker service list on the manager node.

[thaJeztah]

@dperny ptal