Handle the case where the handler fails to self healing an action

We should in case the alert-manager fails to run a command etc,

in this case we need to fire back an alert to prometheus server as "critical" even higher , so that an operator can handle this failure.