Implement support for starting jobs without a queue system

[Nicklas Nordborg]

This would be a feature to make it easier for developers to implement and test new scripts and pipelines. Instead of going through an OpenGrid or Slurm queue manager a dedicated development machine (this could for example be a virtual machine) could be used to start the job scripts directly. The standard commands could be replaced with something like:

• qsub <script> --> nohup <script> &
• qstat and qacct --> ps and maybe read from a special status file
• qdel --> kill -- -<taskid>

[Nicklas Nordborg]

In 6613:

References #1369: Implement support for starting jobs without a queue system

Added the DIRECT cluster type. The engine is currently a copy of the Open grid engine.

[Nicklas Nordborg]

In 6614:

References #1369: Implement support for starting jobs without a queue system

Submitting a job should now work as well as getting basic status information.

[Nicklas Nordborg]

In 6615:

References #1369: Implement support for starting jobs without a queue system

Added another wrapper script run.sh for starting, error handling
and cleanup while running the actual job.

Aborting a job has been implemented.

It should be good enough for development and testing.

[Nicklas Nordborg]

In 6647:

References #1369: Implement support for starting jobs without a queue system

Fixes an issue with VirtualBox and windows shared folders.

[Nicklas Nordborg]

In 6688:

References #1369: Implement support for starting jobs without a queue system

Updated documentation in the example configuration file.