Add support for health checking down after receiving a signal or a filesystem update

[kriswehner]

Right now the only way to proactively healthcheck down the changeserver or snapshotserver is by curling an endpoint.

We need to be able to

• send a (configurable) signal that initiates a healthcheck down
• watch a (configurable) filesystem location that, if the file exists, we healthcheck down, and when removed, we healthcheck back up. This enables proactive maintenance.

[davent]

Currently looking at this