Moved the jump destination after not scheduling to take up/down status into account on next iteration.

The new configuration parameter has a compile time defaul (previous did not).

Unified and extended the logging.
Added end of session logs.

The detection is now done when creating the agent entry,
which happen before constructing the scheduler.

Also improved some exceptions for to-be implemented functions.

Found in case where by default docker does not install documentation files.

This reverts commit 341d1c7f.

After much discussion the commit being reverted has and would have
generated more work than problems solved.

Before this commit, Daniele had hardcoded into the CTA
front-end the constraint that CTA admins must be
authenticated by krb5.

This commit drops this contraint for two reasons:

1. The xrootd/EOS configuration files already require
   operators to specify which authentication methods can
   be used.  Daniele’s constraint is not needed to
   further enforce security, it only serves to constrain
   admin authentication to krb5 which I personally do not
   see as providing any added value.

2. The CERN infrastructure uses krb5, other sites like
   RAL do not.

The destructor of the child process was wrongly de-registering the signal handler from
epoll. This is now prevented by calling the extra post-fork cleanup function for handlers
not involved in a child process before actually running the child process itself.

1d92302d5304266ee14a86b8d9fcbd671b567e5c
  Fix for drive::xx::clearCompressionStats
    - For DriveT10000, DriveMHVTL: stateful accumulator
    - For DriveIBM3592, DriveLTO: log select on specific log page

e6780a8f0a5ed6366ae750eaecc189ac7d0d07fe
  Fix overflowing mountTotal{Read,Write}BytesProcessed by bumping up
  to 64-bit container

12c31edf5193dae90ef65eaa6c3d0eb81a7c6927
  Refactor fix for drive::xx::clearCompressionStats

f626a773aae60b36146d3170e4eb2ef5fd13db35
  Merge branch 'fix_clear_compression_stats' into 'master'
  Fix clear compression stats

  ## Description
    The aim of this merge request is to fix the
    `drive::clearCompressionStats()` method which deleted all log metrics
    with log select on page 0x00 (aka reset all metrics).
    The following changes are introduced:
      * For **IBM and LTO drives**: We are selectively clearing only the
      * compression statistics (pages 0x38, 0x32 equivalently)
      * For **Oracle and MHVTL drives**: Wa are making compression
      * statistics stateful, by accumulating values and subtracting on
      * each flush
    Moreover, we are piggybacking a fix for overflowing
    mountTotal{Read,Write}BytesProcessed
    (drive::xx::getTape{Read,Write}Errors) by bumping up the container map
    in 64-bit value

  ## Testing
    The results of the changes were tested against IBM and Oracle
    drives.
    More specifically, we attempted to write 40 randomly generated files
    of 2GB in size with the previous CASTOR version (2.1.16-9), extracted
    the compression metrics and compared

  ### IBM
    **before:**
      ```
      Oct  6 12:23:09 tpsrv240 tapeserverd[14257]: LVL="Info" TID="14268"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tps
      Oct  6 12:24:53 tpsrv240 tapeserverd[14257]: LVL="Info" TID="14268"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tps
      Oct  6 12:25:45 tpsrv240 tapeserverd[14257]: LVL="Info" TID="14268"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tps
      ```
    **after: **
      ```
      Oct  6 12:34:28 tpsrv240 tapeserverd[14836]: LVL="Info" TID="14847"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tps
      Oct  6 12:36:12 tpsrv240 tapeserverd[14836]: LVL="Info" TID="14847"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tps
      Oct  6 12:37:04 tpsrv240 tapeserverd[14836]: LVL="Info" TID="14847"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tps
      ```

  ### Oracle
    **before:**
      ```
      Oct  6 12:21:44 tpsrv614 tapeserverd[3062]: LVL="Info" TID="3073"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tpsrv
      Oct  6 12:24:02 tpsrv614 tapeserverd[3062]: LVL="Info" TID="3073"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tpsrv
      Oct  6 12:25:12 tpsrv614 tapeserverd[3062]: LVL="Info" TID="3073"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tpsrv
      ```
    **after:**
      ```
      Oct  6 12:34:58 tpsrv614 tapeserverd[3691]: LVL="Info" TID="3702"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tpsrv
      Oct  6 12:37:16 tpsrv614 tapeserverd[3691]: LVL="Info" TID="3702"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tpsrv
      Oct  6 12:38:26 tpsrv614 tapeserverd[3691]: LVL="Info" TID="3702"
      MSG="Reported to the client that a batch of file was written on tape"
      thread="ReportPacker" clientHost="tpsrv
      ```
See merge request !11

Remove debug printouts as the previous commit solves the race condition in CI we were investigating.

In recalls, failing any step does not require stopping the mount nor
end the reporting so we can try and report everything until the session
is ended (by an expicit end session report).