second lexer reg memory use issue

[zzzeek]

>>> from mako.lexer import Lexer
>>> template = '<%def name="foo()"> this is a def. </%' + ' ' * 10000
>>> Lexer(template).parse()

[sqla-tester]

Mike Bayer has proposed a fix for this issue in the main branch:

replace "dot" with "set not containing whitespace" https://gerrit.sqlalchemy.org/c/sqlalchemy/mako/+/4091