You're right, of course. The point was glibly making was that Unicode has a lot of stuff in it, and you're not necessarily stomping on someone's ability to communicate by removing part if it.

I'm also concerned by having to normalize representations that use combining character etc. but I will add that there are assumptions that you can break just by including weird charsets.

For example the space character in Ogham, U+1680 is considered whitespace, but may not be invisible, ultimately because of the mechanics of writing something that's like the branches coming off a tree though carved around a large stone. That might be annoying to think about when you're designing a login page.