Collector instances deployed by the operator expose internal metrics. These metrics should be collected. 

There are probably a couple of ways how these metrics can be collected - using the prometheus exporter or OTLP.

If Prometheus exporter is used the operator should create service/pod monitors.

Acceptance criteria:

• a design document outlining the current state and proposed solution. The doc should take into account that the current /metrics endpoint might change to OTLP in the future.
• share the design doc with the community via a Github issue on the operator repo
• discuss the proposal on the operator SIG call