No need to use weird characters. Greek or Cyrillic letters are enough.
Suppose I download a library with language modules such as Русский.py or Ελληνικά.py; they are allowed as identifiers and can be regularly imported... on utf8 system at least.

Actually such a project already exists: https://code.google.com/p/hellenic-language-toolkit/
"svn co" will correctly create files (win32 explorer show correct names); when importing from IDLE, I get 

>>> import HLT
Traceback (most recent call last):
  File "<pyshell#0>", line 1, in <module>
    import HLT
  File ".\HLT.py", line 29, in <module>
    from Ελληνικά.Ελληνικά import Ελληνικά
UnicodeEncodeError: 'mbcs' codec can't encode characters in position 0--1: invalid character