plus a few cosmetic changes.
Also adapted the corresponding man pages.

Fix following option 2, i.e. double check the Nameserver, for
Felix to deploy at ASGC immediately.

to make the creation script work again.

as OK to be dropped. This state will eventually disappear
but for now it needs to be taken into account.

the generated script on commit r24829

and removed the last 'die-hard' entries of site.def by hardcoding the list of
secure protocols (that is, KRB5) as a define in a header file.
Note that there's still other code depending on compile-time defaults
injected by Makefiles.

so that it's there in all cases.
Dropped sysconfig for security as already not supported/obsoleted.

Fixed nasty error reporting when trying to cleanup diskservers after a transfer removal and the connection was lost. We ignore the error and leave the synchronization to do the job later

Fixed broken initialization of the diskmanagerd queue after the refactoring introducing objects instead of tuples

Fixed cleanup of stager DB for broken transfers on diskmanagerd restart : gets and updates were not considered

Fixed error reporting in case of failed transfers by using put/getUpdateFailedProcExt rather than the non Ext version

Run syncrhonization between transfermanager and DB on all files older than 5mn rather than only files older than 1h

      - for the standard worker one has to add the --xrd option to the comand line
      - for the random and racecond workers the xrootd read/write operations are
        added by default

At the same time the use of a ResourceHelper has been dropped and the d2dtransfer and stageJob modified to take directly the diskserver and filesystem on the command line, rather than getting it from a file given on the command line.

connecting to the DB, as in most head nodes such a direct connection
cannot be established.

The direction is to have only commented options in the example file distributed with only 2 types of options : the ones with a valid default and the ones where there is no default but the associated daemon won't start if it's not present

with xrootd 3.3.1-1

Updated the tape HOWTO to show when the main thread of rtcpd gets the full request from the client.  The getting of the request is before the creation of the pool of disk I/O threads, this means that we could easily modify the code of the rtpcd daemon to support a different number of disk I/O threads when migrating and when recalling.  We would effectively replace the RTCPD THREAD_POOL configuration parameter with two new ones, for example RTCPD RECALL_THREAD_POOL and RTCPD MIGRATION_THREAD_POOL.

        parameter was used. Related to: https://savannah.cern.ch/support/?136498
        Drop the script used to do the xrootd git checkout

SvcMapping and DynamicLib sections, which are now commented out
in castor.conf.example. This means all daemons are linked against
libcastorcommonora at link time, so that dlopening is not needed.
Also provided more verbose examples for the HOST entries.