Prevent edge case when partial archive archives and latest archive is invalidated (should generally not happen since in that case it would be processed before the partial archive, but just in case), by removing the method. The removal is done in ArchiveWriter as well anyway, so it shouldn't be needed here too.
Also another small potential edge case I saw in
applyScheduledReArchiving() if for some reason it is ever called w/o superuser access.
fyi @diosmosis few tests are failing, not sure if due to this PR
@tsteur found another issue while fixing tests: when invalidating partial archives we don't want to change the done value for DONE_PARTIAL archives, since that will also change the semantics. Ie, DONE_INVALIDATED archives are in some cases treated like DONE_OK ones, but DONE_PARTIAL ones should never be.