MDSMonitor: raise health warning when there are no standbys but there should be
We could have a per-filesystem setting for how many standbys it wants to be available. Then if the available set of standbys doesn't satisfy that, raise a health warning.
By default this setting would be zero (i.e. if you just start one MDS ever, that's OK, no health warning), but we could auto-bump it to 1 if they ever ran a standby. That way, if a user started a typical 2-node system, and their standby died, we would raise a health warning instead of being silent.
If anyone wants to silence the health warning they can always set this back down to zero.