Currently the Harvester app will not abort a harvest when Trove throttles the request (due to a missing 'key' parameter); instead it will blithely harvest the <c:error/> errors documents returned by the TroveProxy.
[ ] TroveProxy should report upstream errors nicely.
[ ] Harvester should abort if an HTTP request fails, and report the error.
Currently the Harvester app will not abort a harvest when Trove throttles the request (due to a missing 'key' parameter); instead it will blithely harvest the
<c:error/>
errors documents returned by the TroveProxy.