Introduce concepts to test different schedulers and their versions

[espenfl]

When we in the future become more dependent on scheduling parsing etc. it makes sense to introduce a testing framework such that we can, say the output from stdout/stderr and the detailed job info for the different schedulers and their version.

[espenfl]

I believe it would also be very nice if we could test what happens when we actually go past the physical available memory, or borderline cases. In order to avoid crashing computers or destroying testing frameworks it will probably be useful to execute some kind of vm where a set memory limit is present.