Pull out prometheus network receive error alerts for OPNsense to exclude
its WireGuard interfaces, which like to throw errors, but which aren't
of importance.
Differentiate by host task (hypervisor or not) and disk (hard disk or
not) type not by whether or not the host is physical and virtual and
then by disk type.
This is in line with the disk rate alerts changes and allows for
fine-grained adjustments based on the host task type, which actually
matters for these alerts.
Not by a mix of host task type (CI server or not) and whether or not the
host is virtual or physical.
Also only differentiate on the duration not the rate, to not
accidentally exclude slow hard disks.
This groups the files and templates for each host together and therefore
makes it easier to see all the (config) files for a host.
Also clean up incorrect, unused docker_compose config for mumble and
clean up unused engelsystem configs.
