Unfortunately, and I'm just going to plain say it outloud, @Kevin's answer is actually wrong, but only by a tiny bit.
As @Kevin originally, and incorrectly, assumed was that the rook had to be captured with checkmate. While that is generally true for most rambling rook puzzles (In fact, Tim Krabbe himself invented the term "rambling rook!), it turns that ...
At risk of being wrong because of a mental mistake:
It's easy to reason that the total amount of moves by White must be even:
our first Black moves are
at which point we can finally lose our tempo. Since White's next move is 28. Rd1,
for a total of
As per requested by @Dorian Fusco, here is what my original solution was before. @BoredAtWork, amazingly, and to my surprise, found an alternate way! Congratulations to them!
My Original Solution:
Now do you see why I considered this to be a tricky question? ;D
Beginning of answer (?)
This time around, the pawns are the obstacle, as they can move forward to defend all while being promoted
What a cruel little sister though, she could have ended his misery with just one move starting with this board !
Well, I'd thought to prove that it was insoluble
...and this is the point where my proof falls apart and I realize that @Braegh has the right of it because...
I've looked at this problem for so long that I begin to think perhaps the composer made a slight error in the composition. So I think it would be smart to ask the composer if my try is what he had in mind before I continue cracking my brain. My try is as follows: 1.Bh5+ Rg4 Then the next 4 black pawn moves are forced (irrespective of the order in which they ...