macOS already does different encoding for filenames in Japanese than what Windows/Linux do, and I'm sure someone mentioned same situation in Korean here.

Unicode is already a non-deterministic mess.

And that justifies making it an even more complete mess, in new and dramatically worse ways?