Point made. Well done. It seems that the D and R columns and lines ought to be 50% each for the issue to null and void.
That wouldn’t matter so much.
What matters is not putting the % of the total population on the same scale as the % each party wins within a population subgroup. That’s what is comparing two different things.