[RESEACH] Add comprehensive metrics and tracing

[cupofcat]

First, there is some research to do. We must identify the question we want to answer here

• how many clients are listening?
• how many flags have been evaluated?
• how many evaluations are erroring?
• sync source update data/metrics, when were they last updated?
• latency/duration of flag evaluations
• what flags-sets/sources/versions are available
• what selectors are used by client
• propagation time from source update to client notification

Once we understand these, we need to build metrics for them. This may be done as sub issues.

Relates to: #1141

[aepfli]

#1692 might be also relevant or at least an interesting idea here - with a test implementation here #1595