The monitoring unit HelmRelease does not directly produce the prometheus StatefulSet. It produces the intermediate `kind: Prometheus` which is the one creating the prometheus StatefulSet. Same for prometheus alertmanager StatefulSet which is produced by a `kind: alertmanager` resource. To have a proper readiness check be done for Prometheus, we could introduce an extra monitoring-ready unit, as a dummy unit adding a `healthCheck` on these two `kind: Prometheus` and `kind: alertmanager` resources. (with Flux 2.8 -- recently released, see !7033 -- we should be able to have this be done by Flux HelmRelease controller, without the need for any extra unit)