Follow-up of #2016.
tl,dr: It would be to add the analysis tools of benchmark results to the google-benchmark Python package instead of keeping them in a separate directory. Benefits of this are easier installation for the user, being able to use them outside of CMake, and better organization of Python sources for GBM.
Bonus: The absl-py dependency should be dropped. Since that is used for flag parsing, bindings for the flag values should be added as well as an argparse solution.