Common "text code editors" don't understand a thing about the code they edit. They're just supplemented with ton of hacks that produce meaningful result in acceptably high number of cases (like regex-based highlighters) but still don't have "real" intelligence towards the content of their buffers.
And with ever-growing number of hacks the thing's unlikely to be small or low on dependencies. You'll end up with another emacs/vim/etc - a fairly small cores, and then a scripting or external tool execution support to do highlighting, parsing, context-aware searching, refactoring etc etc to make them useful for code editing. (And then why bother when we have emacs and vim already?)
That's acceptable as long as it works.
And, well, if they don't work, it's certainly easier to patch their hacks or pile even more hacks, rather than trying to redo everything from a clean slate.
Some need to compile the program to work. (So they can't work if the code has compile time errors) Some still don't retrieve all the correct candidates for autocompletion.
Also, it's been awhile since I've tried a clang based plugin but I can't recall any of them controlling the syntax highlighting (eg. Show false #ifdef sections similar to comments) or providing a call tree or jump to caller feature.