You are the one cherry picking perhaps the first two charts. Look at the various metrics at the end in the Zstd vs Pandoc charts. There is just no clear advantage for Python across those metrics which are the one you really want to know about! If anything F# seems like a better top of the line language?!
I'm exclusively using the third chart. The selection process would naively be y=mx+b with some reasonable variables. It would be a downward diagonal (you would move it right until only a few remained), and you would pick the one farthest from the line, which is obviously Python.
