Add PDBs that cover most Pods
Closes #341 (closed)
Mostly this is pretty standard 1-pod disruption budget for handling node failures and rolling updates. We'll obviously need to bump replicas (where it is possible) as a next step to really get HA out of this.
I left out the runner because I'm not sure how gracefully stopping runners would be handled. Is that one that can't be stopped?
Edited by Corey O'Brien