Fixing repos priorities: as we are now pulling from the TAG repo we may have more recent EOS version in the CTA-CI repo (this was not the case for the commit repo). EOS tag repo has now a higher priority and yum cache is purged.

cta-fst-gcd now reports every 10,000 files that it has considered them.  This should help us see how fast the garbage collector is

Changed deafult release from 0 to 1 because 0 should never be used, see https://fedoraproject.org/wiki/Packaging:Versioning

Fixed.

Fixed

Enabling EOS TAG repo for nighly regression tests: if these fail we can now complain to the EOS team.

Upgraded to EOS 4.3.8-20180821144441gitbf49ffa in order to get the FST that does not fail an end-user close() when there is an error with the CTA frontend

Reverted commit of submodule xrootd-ssi-protobuf-interface to b92937743ef5f2eaf1e30336519c0e7ee055e6b2 in order to make the CI green again

- timeout limit increased to 600s
- drive status is no longer set to Unknown
- only effect of exceeding the timeout limit is to display the word
  STALE beside the update time

USerError will prevent the frontend from logging a stack trace.
File path will allow easier identification of problem files.

The tape write task not lister anymore to the memory manager for zero length files.

A retry might sneak into the session (depending of end of sessions and reporting speeds).
Relaxed the test criteria for this case.

Done.

We now leave a placeholder on the tape and carry on in this case. The errors will eventually clear the
archive request.

The disk read task hangs in this case. The tape write taks is still created, will leave a placeholder on the
tape (to be implemented) and count an error for the request.

This reverts commit 5cda9ec6.

This reverts commit a17d1651.