Removed the member variable XrdCtaFile::error that was incorrectly
shadowing XrdSfsFile::error.

Added the XrdCtaFile::m_listingArchiveFiles member variable and
factorized XrdCtaFile::readFromCmdlineOutput() out from
XrdCtaFile::read() so that in the future the XrdCtaFile::read()
method will be able to treat the listing of archive files
differently from other cta commands.

Currently none of the cta commands are incrementally generating
and streaming large result sets back to the cta command-line
tool.  All commands are generating the entire result in the
constructor of XrdCtaFile.  The result is streamed back to the
cta command-line tool, but the result is generated in one
single stepi before any streaming starts.

The listing of archive files is being gradually modified to
incrementaly generate the result in each call to
XrdCtaFile::read().  This commit is one step in this
larger modification.

Put back the serialising of the command return code.  This was
once done within the XrdCtaFile::getMmap() method.  It is now
done within the XrdCtaFile::logRequestAndSetCmdlineResult()
method.

Please note that command return codes must be in the range "0"
to "9" because the first ASCII character sent back from the
CTA front-end to the cta command-line tool is the return code.
The following code snippet from the implementation of the cta
command-line tool shows the format of the reply from the CTA
front-end to the cta command-line tool:

  // The cta frontend return code is the first char of the answer
  {
    uint64_t readOffset = 0;
    uint32_t bytesRead = 0;
    char rc_char = '0';
    const XrdCl::XRootDStatus readStatus = xrootFile.Read(readOffset, 1, &rc_char, bytesRead);
    if(!readStatus.IsOK()) {
      throw std::runtime_error(std::string("Failed to read first byte from ") + cmdPath + ": " +
        readStatus.ToStr());
    }
    if(bytesRead != 1) {
      throw std::runtime_error(std::string("Failed to read first byte from ") + cmdPath +
        ": Expected to read exactly 1 byte, actually read " +
        std::to_string((long long unsigned int)bytesRead) + " bytes");
    }
    rc = rc_char - '0';
  }

Removed unnecessary member variable XrdCtaFile::m_cmdlineReturnCode

Removed unused int return from logRequestAndSetCmdlineResult().

Simplified the XrdCtaFile class.  logRequestAndSetCmdlineResult()i
is now only called by XrdCtaFile::open().

Revert "Revert "cta/CTA#139 cta af ls --vid XXXXX command times out if there are many files on a tape""

This reverts commit 460b9fdd.

This reverts commit e30077f9.

This revert is being made because the original commit broke the
"cta logicallibrary add" command.

The CTA front-ends code for replying to requests is currently
incompatible with streaming back large reply messages such as
the contents of a tape.

Executing the "cta" command-line tool results in the CTA
front-end creating a temporary server side representation of
a file.  The XrdCtaFile class implements this server side
representation of a file.

The constructor of the XrdCtaFile class generates the entire
contents of the file in one shot.  The contents of the file
are in fact the message a user of the "cta" command-line
eventually sees on their screen (standard out).

The entire single shot contents of the file are presented to
the XRootD framework as a memory mapped file which is in
fact a simple "one-shot" array of bytes.  This memory mapped
file approach is implemented by the XrdCtaFile::getMmap()
method returning SFS_OK and setting the Addr and Size
variables.

This commit replaces the memory mapped file approach of
XrdCtaFile::getMmap() with the stream interface approach of
XrdCtaFile::read().

Please note that the CTA front-end is still not actually
streaming the result of "cta af ls --vid XXXX".  This
commit is one step closer.  The whole reply message is
still generated in one shot by the constructor of
XrdCtaFile.  The contents are however now streamed off
to the "cta" command-line tool via multiple calls to
XrdCtaFile::read().

Factorised out the RdbmsArchiveFileItor class so it is easier to
work with and modify.

This branch is a workaround as buildtree runner does not work in master.

This led to VID being marked as available during unmount, and another drive could try an mount the same tape.
Added various logs.

The following sequence was observed in tests (with a sleep(3) at the end of MigrationReportPacker::ReportDriveStatus::execute() ):
 [eric@localhost ~]$ date; kubectl -n cta exec ctacli -- cta dr ls VDSTK11
Thu Jun 22 11:19:10 CEST 2017
    drive   host  library  mountType    status  desiredUp  forceDown     vid
  VDSTK11  tpsrv  VLSTK10    Archive  Mounting         UP             V01001
[eric@localhost ~]$ date; kubectl -n cta exec ctacli -- cta dr ls VDSTK11
Thu Jun 22 11:19:12 CEST 2017
    drive   host  library  mountType       status  desiredUp  forceDown     vid
  VDSTK11  tpsrv  VLSTK10    Archive  Transfering         UP             V01001
[eric@localhost ~]$ date; kubectl -n cta exec ctacli -- cta dr ls VDSTK11
Thu Jun 22 11:19:16 CEST 2017
    drive   host  library  mountType      status  desiredUp  forceDown     vid
  VDSTK11  tpsrv  VLSTK10    Archive  CleaningUp         UP             V01001
[eric@localhost ~]$ date; kubectl -n cta exec ctacli -- cta dr ls VDSTK11
Thu Jun 22 11:19:18 CEST 2017
    drive   host  library  mountType     status  desiredUp  forceDown     vid
  VDSTK11  tpsrv  VLSTK10    Archive  Unloading         UP             V01001
[eric@localhost ~]$ date; kubectl -n cta exec ctacli -- cta dr ls VDSTK11
Thu Jun 22 11:19:22 CEST 2017
    drive   host  library  mountType      status  desiredUp  forceDown     vid
  VDSTK11  tpsrv  VLSTK10    Archive  Unmounting         UP             V01001
[eric@localhost ~]$ date; kubectl -n cta exec ctacli -- cta dr ls VDSTK11
Thu Jun 22 11:19:25 CEST 2017
    drive   host  library  mountType  status  desiredUp  forceDown  vid
  VDSTK11  tpsrv  VLSTK10    NoMount      Up         UP               -

This commit makes the "eos rm" command idempotent with respect to
the CTA catalogue.  If the "eos rm" command requests the CTA
catalogue to delete a file that does not exist with the CTA
catalogue, then this condition is no longer treated as an error.

This branch is a workaround as buildtree runner does not work in master.

Also added extra logs in ArchiveMount::getNextJobBatch().

Just use -D and -O options.

tests/simple_client_ar.sh is the previous simple archive retrieve test, tests/client_ar.sh allows to enter more parameters: number of files for test, their size and the number of parallel jobs to trigger the retrieve.

fixing rmcd log issue as /var/log/castor/rmcd_legacy.log was a named pipe and nothing was collected in artifacts.

and generating all the required krb5 keytabs

using root instead...

Only eosusers and powerusers can write to it, and powerusers only can drop replicas for all file in this directory.
poweruser1 now drops the disk replica of the file before it is recalled in `client_ar.sh` test script.

adding all the missing bits to allow the simple systemtest to run on the client pod:
- add user `daemon` to sudoer in eos so that we can xrdfs prepare as a `powerusers` user
- add the `powerusers` group mount rule in cta
- add the missing bits in client_test.sh