HPC: Multiple Backends

[Schlevidon]

Summary

• Reimplemented the load balancer making it possible to select different scheduling strategies. At the moment the two options are pure SLURM and HyperQueue.
• Made a simple CLI for specifying the scheduler, port and job submission delay.
• Prepared the code base for implementation of another communication layer between the load balancer and a job script in the future (i.e. network-based communication instead of the shared filesystem).
• Performed some general cleanup and added docs.

Related issues

closes #80

[linusseelinger]

Excellent work, thanks a lot @Schlevidon ! :)

I don't see any obvious issues. Merging now so we have this in time for the modeling week; we can merge potential fixes later on.