Professor Walter Mebane at the University of Michigan has written a paper about this github data entitled "Inappropriate Applications of Benford’s Law
Regularities to Some Data from the 2020 Presidential
Election in the United States."

This is as "expert" as you get.  My interpretation: "Nice try, but no."

http://www-personal.umich.edu/~wmebane/inapB.pdf

Mebane is arguably the premier authority on this topic. In fact, it is likely that the person who created the graphs under discussion knew that Benford's Law could be applied to elections BECAUSE of Mebane.  He is the one that applied it to the Iranian elections to prove fraud.

Here is a link to a syllabus for the class that he teaches on election Fraud at the University of Michigan:
http://www-personal.umich.edu/~wmebane/ps485/ps485_syl/ps485_syl.html


Many people like to cite this criticism of Mebane and his application of the law to elections:
https://www.cambridge.org/core/journals/political-analysis/article/benfords-law-and-the-detection-of-election-fraud/3B1D64E822371C461AF3C61CE91AAF6D

But Mebane has responded to this here and everyone seems to miss it. I read this to say "your simulation isn't adequate to tell me that I can't do what I have already done with real data." but he does admit the utility of using Benford's law is an "open question."
https://electionupdates.caltech.edu/2011/08/23/new-research-on-election-fraud-and-benfords-law/