Gather and expose internal metrics

[agourlay]

A coach process tends to run for a very long time therefore it is important to have an understanding of what it does.

Logs help a bit but do not provide the full picture.

We need a /metrics endpoint exposing the metrics for each drills.

• success count
• failure count
• duration
• network latency

Plus a section for the healthcheck for each host.

As a bonus some data about the client

• connection count
• global latency