Look into alternatives to using regex

[JelleZijlstra]

We're using the regex third-party library to avoid a bug around certain Unicode symbols: see #1047 (PR) and #455 (issue). However, this dependency is costly because it's a C library and compiling it is difficult for some users. We should explore rewriting the relevant parts of the parser to not use a regex so we can avoid this dependency.

[ichard26]

And with the merger of #2654 and #2644 the last thing to do is to remove references to regex in the documentation and setup.py, Pipfile[.lock] 🎉