This adds a module to contain benchmark tests (similar to what I did in GalerkinToolkit, with a dummy benchmark. We can discuss what benchmarks we could already add here.

Also adds a github workflow that runs the benchmarks (on push to main), storing the results in the benchmark_results branch. Note that in GalerkinToolkit I called that branch benchmarks but in PartitionedArrays that appears already to exist. We can rename as desired.

Finally, the workflow copies benchmark results over to ghpages so that they can be linked on the documentation site.