add support for utf8 expressions

[woegster]

by simply not using isalpha() but instead replicating the switch() found in the else path, the parser allows UTF8 expressions
I needed this for my side projects, which uses some german engineering system that uses umlauts

[codeplea]

Thanks for the patch. This is a bigger change than it looks: any byte not in the list starts an identifier, so stray characters like $ or @ become unknown-variable errors instead of syntax errors. Closing, see CONTRIBUTING. TinyExpr is small enough that you can add this yourself in a fork. It only stays that way if I don't implement every request like this one.