adjust number of launched workers/processes
in the merging (and probably other parts) of Xsocs, a certain number (8?) of processes is launched for parallel processing.
Sometimes -- at beamline PCs for instance -- this results in severe memory problems. One could estimate the number of workers used by calculating the uncompressed size of a pscan dataset and compare it against the machines memory. Alternatively, one could allow the user to input the number of cores used.
What do you think? Carsten