Add probe for alert rules

[sed-i]

Enhancement Proposal

When alert rules fail validation on the prometheus side, we have a mismatch between the rules in relation data vs. the rules on disk vs the rules loaded and present in the api/v1/rules endpoint.

We should have a probe that can be readily run in production for confirming alert rules are all propagated or not.

Related: canonical/grafana-agent-operator#213