When preventing castling doesn't compensate for a 960 King’s Gambit

I agree that the castling rules in some Schnargls are more off-putting than they are in 518, so I think the Schnargls which put the rooks on the a- and h-files are a fair compromise (I'd joke "Chess960 Lite", but maybe there's some connotation of "960 is a variant") — people can suss castling in those positions more readily, especially when the kings start on the e-file.

The positions where the king and rooks are set unusually present a novel problem: While an opening pawn sacrifice in 518 is usually justified if the opponent is prevented from castling, that might not bother the opponent at all in some Schnargls.