replace mininag with Prometheus once we reach high availability

we have a minimal nagios implementation living on our primary DNS server (currently nevii) that is quite exotic. it checks a list of hosts defined in a YAML file and pings them regularly. when they have a scheduled reboot, it takes them out of rotation, taking care of avoiding the removal of all hosts and all sorts of other corner cases (presumably).

many parts of this could advantageously be replaced by prometheus, once we have high availability. let's do that in phase C.