IMHO, OCaml's decision to not encode strings in a Unicode encoding by default is the correct one. OCaml treats strings as just bytestrings, leaving the programmer to take care of the specific encoding, with the help of sophisticated and powerful libraries like Camomile.
For an example of what happens when language implementations try to shoehorn in Unicode into their strings, look at JavaScript's UTF-16 strings or this Python example: https://www.reddit.com/r/programming/comments/asi2qo/go_is_a...