I do see a handful of people using non-ASCII identifiers in their code, but that's rare. Much more common is explanatory comments, docstrings, etc. in the local language. To require those to be ASCII would be a non-starter.