Maintenance process should notify EOS when it cannot requeue a request
There have been several ops issues recently where requests were failed and then "lost" by CTA (EOS was not informed of the failure).
This happens when a request ends up queued on a disabled tape. The maintenance process tries to requeue the request, but it always attempts to requeue on the same tape, so the requeuing operation fails. EOS is not notified in this case.
The immediate fix for this is to fix the maintenance process logic, so that if the requeuing of the request fails, EOS is notified.
A more comprehensive fix is detailed in #1170 (moved).