Expose leadership information to nodes

[tillrohrmann]

Some components will need information about which PartitionProcessor is the current leader to work in a distributed setup. For example, the query storage component could benefit from this information when fetching the information from leaders running on different nodes. Also, the ingress when using RPCs to ask the PartitionProcessors for a result of an invocation or to submit an invocation will need this information.

Initially, we could propagate this information from the ClusterController that accumulates it via heartbeats. As part of the heartbeats we could send the latest known ClusterState which contains information about the current leaders.

In the future, we want to use a gossip based protocol to propagate the cluster state throughout the cluster.

[tillrohrmann]

Already covered by #1907.