Resolve Erratum 2235 with reference to Unicode, not UTF-8
https://www.rfc-editor.org/errata/eid2235 observes that we don't clearly specify what the hyphen delimiter means within a regex range designator.
Make it clear that we are just doing a Unicode codepoint sort order, regardless of the UTF-8 encoding.
This is intended as an alternative to !330 (closed)