Add probing end point

[steeling]

End to end tests are great for qualifying binaries and discovering bugs, but how do you test for liveness in production?

Probing! By adding an entirely separate set of endpoints /v1/probe/dss/...
To support gRPC probing we would need to duplicate and rename the service, ie: service ProbeDSService {...}

Note this is not part of the spec, but an implementation detail.

We definitely want to keep data separate, so the big question is if we would want to use separate tables or a separate database. I'd lean towards a separate database run within the same CRDB cluster, although that could potentially miss some cases where we bork the database (potentially due to a bad db migration push?). I think this is a small case to catch for larger wins in data separation. If the db is really messed up we will probably detect fairly quickly through traditional monitoring