Implement a backlist for OSDs ( possibly also for other services ? )

[jschmid1]

Description of Issue/Question

When you find yourself in a forensic situation where one needs to figure out why a certain service is failing, one often does not want to remove a service/device just yet. You will want to be able to operate on your cluster without running into timeouts.

There should be a way to tell Deepsea that you know that certain processes are down and you do not care. That might be implemented with a blacklist that lives in the pillar.