1st stable release of haddock-pilot

This contains a simple example of a pilot machinery for running HADDOCK on a HPC system, using multiple nodes. One pilot is started on each node, which runs as long as there is work to be done.

The pilot monitors the 01-TODO directory for gzipped tar archives (.tgz) which are ready to execute HADDOCK runs. When work is found, the pilot will move the archive to the 02-RUNNING directory, touch in that directory a corresponding .tgz.process file to tell other pilots that this particular run is already being handled. The pilot then copies and unpacks the HADDOCK run into the /tmp directory and executes HADDOCK, running HADDOCK in node mode (i.e. the HADDOCK python process is started locally on the node, and the queue command defined in run.cns is simply csh (a version of HADDOCK configured to this end should thus be used). To use a full node, the number of queues in run.cns should be set to the number of available cores/threads on the node.

All computations thus occur locally on the node as the run directory should have been setup with relative paths (i.e. ./). Upon completion of the HADDOCK run, the pilot will archive the run, move it to the 04-RESULTS directory and delete the local directory under /tmp. The input archive is moved from 02-RUNNING to 03-DONE and the .process file is deleted.

The pilot will then look for the next workload. When no workloads are present anymore the pilot will stop.