Fixed a typo in a variable name.

Increated the width of the size column of the result of
"cta archivefile ls" to 12 characters.  The column is in bytes
which means that we can now display a file size of just under
1 Terabyte.

those were 15kB -> 150kB will be better...

RdbmsCatalogue::reclaimTape() now resets the DATA_IN_BYTES column
of the TAPE table.

Improved unit tests to test for the fact that reclaim should
reset the tape occupancy field.

The following unit now "succesfully" fails:

[ RUN      ] InMemory/cta_catalogue_CatalogueTest.reclaimTape_full_lastFSeq_1_no_tape_files/0
/home/smurray/CTA/catalogue/CatalogueTest.cpp:7029: Failure
Value of: tape.dataOnTapeInBytes
  Actual: 1
Expected: 0
[  FAILED  ] InMemory/cta_catalogue_CatalogueTest.reclaimTape_full_lastFSeq_1_no_tape_files/0, where GetParam() = 0x7fc58e93eab0 (7 ms)

The next step is to fix the bug in reclaim.

pushing a bit harder in the tests with 1000 files of 15kB that should fill 1.5 tapes (100MB each in CI).

Enable it in the CI tests.

The ArchiveRequest was not garbage collected if had already been requeued before, as the
corresponding status was overlooked.

Created an objectstore utility for requeuing ArchiveRequests after deleting archive or retrieve queues.

This utility is to be used for cleaning up the pre-prod environement following a schema change in the archive queues.
It could support all object types on the long run, but currently only supports ArchiveRequests.

Metadata can be used for short replies (less overhead than a full reply)
or to provide information about the reply which will follow (e.g. number
of records, response size, etc.)

Alerts can be used to signal the client about the server state.

Created an objectstore utility cleaning up the root entry after deleting archive or retrieve queues.

This utility is to be used for cleaning up the pre-prod environement following a schema change in the archive queues.

Protobufs for different message types will now be handled by template
specialization

Like in ArchiveQueue, added per file cache of parameters, summary rebuilds and rebuild count.

If not, eos would refuse to use the space after finding the .eosfsid file.

Fixed missing decrement of size total when dequeueing ArchiveRequest.

Changed ArchiveQueue's schema in object store so the per-job values are cached locally in the queue.
Added a sanity check on commit() so the summary are rebuilt if incoherent.

This object (with thread) already reports data movement to the main process, and will
also store it in the object store (interface will be added to TapeMount.

Returned some formatting to the output of the "cta archivefile ls" command

Changed the type of the ListArchiveFilesCmd::m_readBuffer
member-variable from std::string to std::ostringstream. This
modification will make it much easier to add output
formatting for the "cta archivefile ls" command in the
future.

Moved the path column of the output of the "cta archivefile ls"
command to the very end of the line.  This was done because the
path column is the only real random width column.

The "cta af ls" command now sends several rows back to the client
in a single reply.

Set the header colour of the "cta af ls -h" command back to red

Everything is now merged in the standard `ctafrontend` docker tree completely removing file duplication.

The clean solution would have been to add cta to the `tape` group.
Unfortunately this does not work as groups are not refreshed within a bash process.
Therefore the short answer was to `chmod 666` SCSI devices (`sg` `st` and `nst`).
Now everything works fine in buildtree but this step could be improved.