Add metrics documentation

[black-adder]

Jaeger collectors/clients emit metrics about the health of the system but they aren't documented anywhere. By documenting them, new users can see which metrics to look out for when setting up jaeger.

[yurishkuro]

Could we auto-generate documentation, maybe? I did want to normalize the metric patterns to be less path-based and more tags-based (esp. the collector metrics).

[otisg]

Jaeger collectors/clients emit metrics about the health of the system

Which metrics is this referring to? Are these metrics about Jaeger, or metrics that Jarger (client) collects about the app it's tracing?

[jpkrohling]

It's internal metrics, to assess the state of Jaeger itself.