I don't understand the argument for "characters = grapheme clusters". From my perspective, there are a lot of different ways you'd want to iterate over a string. Grapheme cluster breaks,
tailored grapheme cluster breaks, word breaks, line breaks, code points, code units… all of these make sense in some context. However, there are precious few times that I've wanted to iterate over grapheme clusters, so telling people that they should do that instead of something else doesn't make sense to me. (I mean, what problem is so common that we would want to iterate this way by default?)
For parsing, it often makes sense to iterate over code points or code units, since many languages are defined in terms of code points (and you can translate that to code units, for performance). XML 1.1, JavaScript, Haskell, etc... many languages are defined in terms of the underlying code points and their character classes in the Unicode standard. JSON and XML 1.0 are not everything.