[WIP] Proposal: ProbeVolumes RPC for volume-level health checks

[jdef]

The idea is that ProbeNode is not the right place to implement per-volume checks and that it may be useful for a CO to assess the health status of a volume:

• prior to pairing it with a workload
• on an intermittent basis whether or not a workload is using it
• .. probably other use cases, flesh this out

ProbeVolumes is a call dedicated to reporting volume health. But where does it belong?

• Node service? probably makes sense for local volumes, but not external/remote vols
• Controller service? probably makes sense for external/remote vols but not local vols
• a new VolumeHealth service?

[jdef]

#236