When the Gatekeeper installation status changes (i.e. not installed to installed or the reverse), the governance-policy-framework pod becomes unhealthy and gets restarted by Kubernetes so that on startup, the gatekeeper-constraint-status-sync controller can be started.

This makes Gatekeeper related tests slower, less reliable, and hard to debug since the pod logs are lost on restart. We should instead, have a goroutine that monitors the Gatekeeper installation state and starts/stops a controller-runtime manager just dedicated to the  gatekeeper-constraint-status-sync controller.

Note that the health endpoint proxy in "startHealthProxy" function needs to account for this new dynamically created endpoint.