Normalization should be handled in the code
Translators should have to provide lists of characters and how to handle them. Unicode normalization should be handled at the code level to make it more robust, accurate and exhaustive. I don’t know enough about your programming language, but there are certainly libraries that handle that.