Remove accents

Strips diacritics and leaves the base letters: "café naïve Zürich" becomes "cafe naive Zurich". Non-letters and spacing are untouched.

This works by decomposing each character into its base letter plus a separate combining mark, then discarding the marks — which is why it handles the whole Unicode range rather than a hand-written list of substitutions. There is one category it cannot help with: letters that are not a base letter plus an accent at all. The Danish ø, the Polish ł and the German ß are distinct letters rather than decorated ones, so they survive unchanged. If you need those transliterated too, that is a language-specific mapping and a different job — and one where the "right" answer differs by language, since German convention turns ü into ue while most others turn it into u.