Check ElasticSearch cluster health in admin dashboard

[renchap]

Pitch

In addition to the version check, we should also check the cluster status (client.cluster.health["status"]) and show a warning/error if it is yellow/red.

A yellow status usually means that replicas are not assigned (one of your nodes are down?), red usually means that shard data is not available and that search is not working properly, or not at all, and your cluster is very unhealthy.

Motivation

I suspect many people deployed ElasticSearch on their instance, but do not have proper monitoring in place and do not know that their cluster is having issues, causing search to either not work, work on only some data, or is very slow.

Note: this should only be done after the recent work on setting the proper count for shards/replica is done, as at the moment every single-node ES cluster would be yellow as Mastodon always sets number_of_replicas to 1.