That is a very beautiful proof game! I find it hard to believe that someone hasn’t covered this territory before. If not, it’s a wonderful idea (a little mechanical, yes, but still…)
Is a faster solution possible, you ask? I wonder whether some of the rook’s Pac-Man behavior (moves 21-25) could be replaced by pawn captures. (Even after pawns are on “correct” files, they could leave those files & come back.)
One might discover something interesting by constructing a proof game for the reflection position (White Kc8, Black Ka8 or Ka7)
Actually, I was sort of expecting that you’d find something in the literature somewhere. I, too, am surprised to find that this is new ground.
Yes, it probably would indeed save a few moves if black were to put most of his pieces on squares the white pieces eventually need to move to anyway. As it is, black is spending most of his time fiddling while white burns.
I wonder if there are positions satisfying the original question where the black king is anywhere other than h8, h7, a8, or a7. Somehow I doubt it.
I may not be able to count correctly but I particularly like the one with the Black king on e8. When I considered blocking in a bishop and putting a knight two square from the king I realized that f8 and e8 could be done that way and those were positions where either player could have moved last and it was still stalemate either way. I also have a version with the White king on c8.
Here’s one with the Black king on d8 and the White king on b7. Ten pawn captures are needed. The last moves were exd6+ Kd8.
With an eleventh pawn capture the d3 pawn can be placed on c5 and the resulting position can be a box-like as 17 pieces can get.
You don’t need the d4 pawn on d2.
White’s last move could have been e6xQd7
If the Black king is already on d8 then Black is stalemated. If the Black king was on e8 then the capture was check and Kd8 stalemates White.