Wrong parameter was passed as retrieve request id.
Added logs to frontend.
Implemented validation of cancel in archive retrive test (no file should be retrieved).

Also improved function name to better match XRootd names.

This can now be done without tearing down and re-creating the containers.

Added error messages in case of problems for easier diagnosis.

The retrieve request address is now stored in the EOS namespace as an extended attribute and deleted
after checks on cancellation. The cleanup of the queues will be eventually done by the tape servers.

Adding a non fatal ERROR for the missing sys.retrieve.req_id (for now) and a fatal one if we intercept some xrootd client failures during the past run

This reverts commit 17abf6f6.

Superseding of tape files by filesWrittenToTape() in now done in FSEQ ACS order in order to cover the corner case of the same tape copy being written more than once in the same archive batch

Google protobuf3 logs a FATAL error if .at() is called for a
non-existent key, so the log is filled with errors if we use try {}
catch(...) {}

Replaced 'COMMIT;' with 'conn.setAutocommitMode(rdbms::AutocommitMode::AUTOCOMMIT_ON)' within OracleCatalogue::filesWrittenToTape()

If CHECKSUM_BLOB is NULL, the code will fall back on CHECKSUM_ADLER32.

[REPACK] Added a failed to create archive request counter so that the Repack Request is complete or failed if Retrieve subrequest is failed (creation or execution)

kill. This is quite likely for cta-taped as its static memory buffer
consume a lot of memory.

Added the new method Rset::isEmpty() - I am guessing that the migration work of Michael would have loved to have had this method