Thread usage is opaque, add monitor on dispatch thread and tooling to understand it

[daviddawson]

Muon Version: 7.1.6
Language: JVM

As seen in muoncore/photon#5 it is possible for protocols to be written that do excessive wsork (even block) the main dispatch thread. This will either slow event dispatch, or cause it to stop completely, killing the muon service.

This behaviour should be monitored and alerted if possible. Other worker pools should also be monitored and metrics provided.

[daviddawson]

Centralised thread usage in 078a189

Now needs the ring dispatch to be monitored and the pool to be managed to fully fix this.