Benchmarking based of the number of CPU instructions

[greymd]

Current benchmarking is calculated based on the processing time.
Rather than that, we should implement benchmarking based on CPU instructions and visualize the result, to avoid changes which may cause performance degration such like #41.

References:
https://pythonspeed.com/articles/consistent-benchmarking-in-ci/
https://www.mizdra.net/entry/2023/02/12/205109

[greymd]

Todo this, GitHub pages must be updated.
Assignee is me since I only have the permission to make changes.