add option for consul_health_node_status to use leader election / semaphores

[chnrxn]

Every node in a Consul cluster will report on every check on every other node. In a cluster with N nodes and C checks, this creates 4*N*N*C which is exponential to N. Querying consul_health_node_status on large clusters can be quite time consuming.

One proposal is to support and optional flag to use a semaphore (size configurable) such that only a few hosts will be exporting consul_health_node_status. This way, we still achieve redundancy without exploding the number of time-series being produced.