Define health-check endpoints for pingdom / monitoring

[dmp42]

The idea is: expose an (multiple?) endpoint(s) that does health checks (on the backend / system / whatever / to be defined) so we can hook pingdom or another system into that.

Need to explore pingdom possibilities first (authentication, custom headers, body processing).

Then need to decide what we want.

cc @discordianfish

[bacongobbler]

Would this be a backwards-incompatible change and we'd need to update our storage drivers?

[dmp42]

@bacongobbler not likely I would like it to be transparent - the default interface would additionally provide a new method that could be overridden (but not necessarily).

Now, this one is more generally about the registry side of things.

About what could be done on driver land, this one is related: #482

[discordianfish]

@dmp42 Not sure if dedicated health checks are the best way, but I guess you can say best. My first approach would be to monitor the same things people actually hit like pulling/pushing things etc.