Add metric for service backend counts

This is almost definitely the wrong place to do this. If you just want a global instance count you should add it to lymph-monitor.

Mar 01 '16 13:03 emulbreh

It is a very good place to put it. Because it will report what the various instances believe to be the case -- and this is what needs to be correct (and hence monitored).

Mar 01 '16 13:03 Drahflow

let's rename it to something that communicates this idea then, known_instance_count maybe, with address as a tag.

Mar 01 '16 13:03 emulbreh