Make the core pod checksum the satellite pods, like it does with core already.

[kuzmik]

Rather than running LOAD PROXYSQL SERVERS TO RUNTIME every Xs, regardless of whether changes occurred to cluster state, it'd make more sense to replicate what we're doing with core pods.

ie: checksum the current cluster state, and every 10s if that checksum changes, run the LOAD command. That will cut down on logs in datadog, at the very least.

[kuzmik]

I actually had the idea to use Informer() to watch cluster state, rather than a manual loop that runs every X seconds. PR is opened and linked to this issue.