Adding R scripts analyzing benchmark results and small test data

[stanisic]

I would like to emphasize that this contribution does not impact any other part of the Ginkgo code. Only in the case of the large changes of the benchmark outputs file structure, these scripts might need certain adaptations.

These 4 R scripts are doing several types of analysis and plotting of the results for the spmv and solver benchmarks. Structure and the reading part of these scripts is following the same approach as the existing tuning_heatmap.R. I have also added small, manually modified, test data that one can use to test scripts and inspect their capabilities.

The author-email-affiliation-license header is just a proposal, please let me know if you would prefer a different approach.

[tcojean]

Thanks Luka for the contribution! Currently, we're busy with a release, we will take a look at this afterward. Other ways to analyze our benchmark data are always interesting :smile:.