Feature: Support HA setup for ClickHouse metrics exporter

[carlos4ndre]

Assuming we want to run the ClickHouse metrics exporter with HA (Highly Availability), one would need to deploy it separately from the ClickHouse operator, which can only run with a single replica.

Currently, state is propagated from the ClickHouse operator to the ClickHouse metrics exporter running within the same pod, but in case the ClickHouse operator is down so will the metrics exporter service.

A possible solution would be to change/extend the ClickHouse metrics exporter logic to do polling or periodically check for updates reusing the logic in DiscoveryWatchedCHIs.

[sunsingerus]

The idea to extract metrics operator as a completely separate entity has been around for some time. However, tasks with higher priority are blocking this notion. We need to continue discussion and gather arguments pro and contra.