Thanks for this.
I am somewhat familiar with Benford's Law, and when I heard that it was "proving fraud", I went "hm?" because it's sort of intuitive that for Benford to be applied, you have to be analyzing a range of numbers that *could* start from 0-9.
But an election......at least in the US, is a trinary. Three possible choices. You voted for X, you didn't vote for X, or you didn't vote for anyone at all.
So, you immediately have a problem with using Benford, over and above what you described, I would think.