Watchdog: Handle HTTP errors more intelligently 

Based on the HTTP scheme, we need to extend totem to more intelligently manage the results. 

i.e.
(404) not found -> push tasking to misbehaved
(403?) API limit exceeded -> push tasking to another queue?
(500) service error -> retask, restart service?
