Documentation gives regex for CLEAN_USERNAMES that isn't the one that's coded

[TeckFA]

The docs say, in Username generation, that the regex for CLEAN_USERNAME is r'[^\w.@+-_]+', whereas the code has (in social.storage.base)

CLEAN_USERNAME_REGEX = re.compile(r'[^\w.@+_-]+', re.UNICODE)

The documented regex creates an interesting range, viz. '+-_'. The actual Python text puts '-' at the end, which is good; also, it has the re.UNICODE option, which, I think, is worth mentioning in the docs, as it provides guidance when thinking about locales for \w.

[omab]

Updated python-social-auth/social-docs@551123c.