Conflicts:
	hsmtools/entersvcclass
	hsmtools/modifysvcclass

Fixed bug #104601: RFE : modify/enterSvcClass should check consistency of the GCPolicies when DiskPools are shared

Conflicts:
	hsmtools/entersvcclass
	hsmtools/modifysvcclass

Fixed bug #104601: RFE : modify/enterSvcClass should check consistency of the GCPolicies when DiskPools are shared

Fixed long options of listtranfers taking an argument. They were broken in their long version as getopt was not aware of the argument

On Put failure, invalidate diskcopy as opposed to dropping it completely so to give a chance to the GC to drop whatever is on disk

	which contacts the diskmanager and passed the status of the operation.
        Drop also the unlink command which was contacting the XRootD daemon on the headnode and
        requesting the deletion of the file from the namespace in case an error occurred during
        the transfer - the cleanup will be done by the diskmanager as it will know from the XRootD
        daemon on the diskserver that the transfer failed.

This includes:
- a new thread to reply to clients using the CASTOR2 protocol (cf.
  stager's RequestReplier).
- a new thread listening to clients coming back to start the transfer.
  This thread polls all outstanding client connections and just accept
  them, so to pass the file descriptor as stdin/out/err to the mover
  (inetd mode).
- a new thread listening to movers for their file close
  operations.
  Details on the implemented protocol in moverhandler.py.
- Extra exposed_ methods to handle the meta-data operations.
  Those replace the stager Job service, which is now obsolete.

Modified gridftp to work as a child of the diskmanagerd instead of a stagerjob. Essentially calling mover_file_close at the end of the transfer.
This is still incomplete - only successful writes are handled correctly.

Modified rfiod to work as child of the diskmanagerd instead of a stagerjob. The non-standalone mode is used, and more cleanup can now be done in the command line arguments as most are obsolete.

Introduced a DiskManager/MoverHandlerPort configuration parameter, and removed the now obsolete second timeout in xroot

Added a small library for the movers to handle the metadata operations when opening and closing files. The location is temporary, the code of all movers will be grouped together later on.
For the time being, only the close procedure is implemented.

Fixed stats done by rtcpd for recalls : in particular, do not try to stat directory in which the file should go when we use an object store !

Added protections against division by zero and inaccurate statistics gathering for actual tape space usage.

We always floor the compressed size to files to one.

Fixed in the master ready for a 2.1.15.x release.