Parallelize benchmark execution

[hidmic]

Feature description

Benchmarks can take a very, very long time when running against long datasets in real time, as it is often necessary to ensure stable system execution and representative results. LAMBKIN does not currently support any form parallelization, when it could trivially do so. We should devise a mechanism for this.

Implementation considerations

Multiple, independent benchmarks running on the same machine will invariably interact: all computational resources are shared, and this will skew results. Safe parallelization will likely require multi-machine support.