Revert "Revert "Removed 'Do nothing' logic from cta::server::Daemon which unwantedly kicks in if systemd is used""

This reverts commit ba2116da.

Reverting the revert because nothing changed.  The GitLab CI still
does not succeeed in running the "init" pod.  The commit I made
and then reverted appears to have nothing to do with the problems
of the GitLab CI.

Revert "Removed 'Do nothing' logic from cta::server::Daemon which unwantedly kicks in if systemd is used"

This reverts commit 462959d4.

The current pipeline in the GitLab CI keeps failing to complete
the "init" pod's work.  The same Docker image is used on
stevedevctai.cern.ch to successfully run systest.sh -n ctatest.

The GitLab CI does not provide any logs from the "init" pod and
therefore cannot be debugged.

This revert is a desperate attempt at fixing ithe problem.  If it
does work then it means the GitLab CI is benefiting from cta-taped
remaining as the root user.

Fixed calls to promise::get_future() after possible access form other thread. They are now guaranteed to happen before.
Added helgrind annotations for promise based synchronisation.
Added macros enabling helgrind annotations for shared_ptr.
Added suppression for shared_ptr used inside other standard lib object and not covered by the previous macros.
Added unit test for lower level .
Added suppressions for reported race conditions in Rados library.
Review heavily MemArchiveQueue and fixed missing commit in object store, leading to potentially orphaned objects.
Enabled formerly disabled test as it is now fast enough.

Changed user stage:st to cta:cta

cta-taped now logs the user and group that it is setting for the
current process.  For example:

May 18 16:54:13 itdssbuild01 cta-taped: <158>2017-05-18T16:54:13.402884+02:00 cta-taped[17210]: LVL="Info" TID="17210" MSG="Setting user name and group name of current process" userName="stage" groupName="st"

Add publish_tagged_rpm and needed helper derived from https://gitlab.cern.ch/ci-tools/ci-web-deployer as we need to run the update_repo.sh hook after the file transfert.

Adding cta_tagged_rpm stage. CTA_RELEASE and CTA_VERSION are not exported as is to avoid overriding environment variables used by cmake. Therefore I am using TAG_VERSION and TAG_RELEASE.

Fixing issue #108: rpmbuild runs cmake underneath, but this second instance needs to propagate the variables passed to the initial cmake.

Fixed.  The cta command-line tool will now timeout after
15 seconds when trying to connect to the CTA front-end
daemons.  The previous timeout was 8 minutes.

The test remains disabled as it still take quite a bit of time.

This is done by replacing unique_ptr/bare pointer with 2 shared_ptr.
Also added suppressions for helgrind complains about rados.

Print CTA_BUILD_ID in the jobs and not the full SHA of the commit of course. The generated rpm version was OK but the logs were useless.

`CI_BUILD_REF` environment variable has been dropped in [gitlab 9.0](https://cern.service-now.com/service-portal/view-outage.do?n=OTG0037271)

Fixed.

Removed the dependency between cta-cli and the xrootd
"meta package".

Lambdas were stored on stack and passed to async. They are now stored in object with proper lifetime.
Improved ArchiveMount::getNextJobBatch(): we now clean up the empty queues.

BackendVFS now throws the same exceptions as BackendRados.
The exceptions of the user provided update callback are now passed through.
A unit test was added.

This reverts commit 22685812.

Added comment to cta_catalogue_db.conf.example that explains which connection string should be used when

Added an intermediate interface to reuse SchedulerDatabaseFactory.
Added optional debug infos for BackendVFS constructor.
The unit test is currently disabled.

The ArchiveJobs are now updated using asynchronous IO.