Feature request: Improve telemetry detail

There was a conversation in HashiDays London, regarding production workload analysis on consul, at the end of which we concluded that consul could report raft metrics, such as 'applied_index' via telemetry endpoint. A reason for that conversation was a constantly altered index.

In addition to that we saw benefit of having a counter per catalog entry to monitor flapping services (each service flap would increase the counter). Since the same idea could be requested for k/v store updates, a more generic approach should be considered.

Few ideas:
1. Allow to turn on/off detailed counters for catalog and k/v domains
2. Allow to define regex filter for k/v domain (catalog too?)

After inspecting existing telemetry measures, I found several things that I would like to see changed:
1. Only elected master should be exposing general raft telemetry (not all servers should expose that)
2. node-originating telemetry should have node name attached to them (similar to current nomad telemetry structure)
3. Telemetry naming should be reviewed, because there are problems with designing general metric name match for statsd protocol, due to information placing in current names.