Add possibility to define number of cpus for tests

When running the Dart.sh script normally all available cores are used which
works well in most cases. Up to now it was possible to limit the number of
used cores by defining the environment variable NCPU which was used for the
build as well as the test stage. Depending on the system this resulted in a
large number of tests which ran into timeouts which were not seen when running
the tests sequentially. With this commit it becomes possible to limit the used
cores during the test stage independently. This is done by defining the
environent variable NCPU4TEST.