Too late to edit, but the situation with İ is more complicated, and I got the mapping wrong for this specific casefold implementation. What I should have said

Too late to edit, but the situation with İ is more complicated, and I got the mapping wrong for this specific casefold implementation. What I should have said is that in the case of İ/i/I/ı, using str::to_lowercase for string matching wouldn’t be any less correct than using their locale-independent casefold. The third character in the list, final sigma, is a good example that illustrates why using str::to_lowercase for string matching isn’t good.