Replaced oracle::occi::setVector(occiStmt.get(), 1, archiveFileIdList, SYS, ODCINUMBERLIST) with temporary table TEMP_TAPE_FILE_BATCH

The retrieve request being queued in one queue only, we can add a reference to
it in the queue, and then create the request object after. This is a simplification
from the previous algorithm.

Added logs for request creation. This will now create a log line where both the object
name and the file Id and path are available as parameters, easing debugging,.

Also corrected some logs.

Also fixed some logs.

OptionStrList takes a filename as its value. It reads the file into a
list of strings in the protobuf. On the server side, the values are
unpacked and made available as std::vector<std::string>.

For example implementation, see "cta-admin repack add"

The jobs will now remain attached to the agent of the drive process. The garbage collector
process will now requeue them instead of drive process. This will allow the drive process
to restart faster for a new session and the drive process to requeue the jobs faster a it
processes jobs in bulk.

We should not be overdoing it: it should be owned by user cta but we do not care about the group. This was failing in CI as cta group does not exist anymore

Also cut the main garbage collection function into smaller bits.
We requeue jobs in batches of 500.

Successful retrieve requests are asynchronously deleted as soon as they
are complete. Then from time to time (every 500), the requests get removed from
process ownership in bulk.