BENCH, ENH: runtime heatmap performance on Python side

[tylerjereddy]

We'll likely prioritize correctness over speed in gh-615 (which makes sense). Moving forward, we may want to add asv benchmarks, possibly with synthetic data until darshan-hpc/darshan-logs#15 is resolved, and check for any low-hanging performance fruit.

I was reminded of this by this comment related to log file e3sm_io_heatmap_and_dxt.darshan (emphasis mine):

I'm no longer seeing any errors and can generate a summary report for that log with using this branch (it does take a couple of minutes).

[tylerjereddy]

We could also compare the DXT vs. runtime heatmap plotting performances there, etc., to see how big the gaps are.