Ported improvements of Helpers::getLockedAndFetchedQueue<ArchiveQueue>() to Helpers::getLockedAndFetchedQueue<RetrieveQueue>().

Also used lockfree in initial access to root.

This reverts commit 3b1855a8.

The 'SELECT FOR UPDATE' was not the bottle neck.  I am putting it
back for now.

session finished" cta-taped log messages always be on the end.

updating wfe.ntx to 10 and wfe.interval to 1 as 3 was way too low (1 is to match preproduction number of wfe.interval).

To activate this option run CMake with:

$ cmake -DBUILD_XROOT_SSI=1 ../CTA

Especially, in OStoreDB::fetchMountInfo(), some operations were not lock free where they should have been.

Proceed all successful reports asynchronously and periodically check
and clear statuses if they have finished.
In the end of session do the check/flush for all reports in the
successful reports queue.

Switch from synchronous rados remove to async aio_remove in
case of zero size object in BackendRados::AsyncUpdater

As those fuctions are usually called in a tight loop at the end of the session or in the garbage collector,
they tend to unlock and relock the queues immediately, and sitting on the lock for a long time in practice.
The added sleep will increase the odds of other processing slipping in between the unlock and the lock.

The ping() and queueRetrieve() timings are of interest, so commit now.
The rest of the timings will follow.

Use precision .00 for Mbytes. Move library name to the front.

This led to failure to requeue non-handled jobs at the end of a tape session (and leave them
to garbage collection.
Also added timings.

Removed all instances of RetrieveMount::getNextJob() and siblings.
Fixed incomplete deserialization of archive file in RetrieveRequest::asyncUpdateOwner().

Done.

Changed reporting to the Catalog with a batch of written files.
Changed synchronous reporting to the backend job by job to the
asynchronous reporting for batch of jobs.
Changed synchronous reporting to the EOS mgm to the asynchronous
reporting.

This new version locks the queues for less time.
Fixed a bug where the wrong lock (shared and not exclusive) was taken when removing empty queues
from the root entry.
Improved multiple iterations retrying (we can now retry on a new queue if one gets contructed
in the meantime.