NEMO: Upgrade of job scheduling software suite

Unfortunately, the upgrade did not go as smoothly as expected. But since we need the new features, we chose to subsequently fix the problems instead of rolling back to the older software suite. Rest assured that we will be more conservative with such changes on the final bwForCluster NEMO. Among the new features is the enforcement of given resource requirements, in particular for memory (RAM). The new mechanism is technically based on the cgroups feature of the Linux kernel and protects users giving proper memory requirements from users who underestimate their job memory requirements. Before, when users where sharing a node, a "good" job could get killed or slowed down by a "bad" job. Now, the bad job is terminated as soon as it tries to allocate more memory than requested in the job definition.

We are glad that we can iron out this category of problems while running on the preliminary cluster and we would like to apologize to the users currently using the test cluster for the inconvenience and thank them for their patience.