And also - why do the mail-in ballot batch ratios stabilize and end up being identical to the final overall state tally, every time? It's almost as if he's plotting just the standard D/R overall ratio instead, except for all the noise at the start.

Something seems off to me as well. The vote ratio should pull harder in a certain direction to actually change the election percentages and shouldn't end up with the same as the end election % result. What these numbers don't show is the volume of the batch however, which would be something else that is valuable to look at.