#339 Look into swapping of Bfast jobs on cluster

[Jordan Mendler]

The idea is to see if we could have 2 bfast jobs running on the same machine, with 1 of the jobs suspended and swapped to disk. This would allow us to use SGE queue subordination, where we can have a long.q (no time limit) for long running jobs, and an all.q (hard time limit of 6, 12 or 24 hours) for normal jobs. As all.q fills up, it would suspend jobs in long.q, flushing long.q's bfast to a large swap space so the system does not run out of memory.

This would allow users to submit lots of long running jobs that do not interfere with production/non-power user jobs, but also can use all idle resources on the cluster.