There are cases where asking for depth>0 node_info will 500. This breaks
listing jobs because we raise and stop looking for running jobs. Handle
this by checking if it is a 500 error at depth=2 and if so check
depth=0. If that returns successfully treat it as a broken slave that
isn't running any jobs.
If depth=0 returns an error things are probably much more broken and
should be looked into.
Change-Id: Ieac15a0fe2a47ec3dae51db96ad2fe40992c353a