Disregard my last post... I see a mistake.
Even if there is a mistake in the math in your last post (if there is one, I didn’t read closely enough to catch it), the point you make is a good one - the ‘count each card’ method of calculating turnout does provide some room for virtually undetectable manipulation, as under votes provide an opportunity to slip in extra votes without looking too suspicious. Frankly, it doesn’t look like that happened here, at least with respect to the Presidental ballots (about 70% of registered voters cast ballots for president, which is consistent with the overall turnout, and I would think that any one-sheet under-votes would likely involve the second sheet, not the one with the Presidential ballot), but the room for manipulation is certainly there.