Remove an now-unused import in the tokenizer.

Two still fail because templates aren't implemented yet, but those
are otherwise handled correctly.

All tag tests passing. Still need to finish backslash support and
support for templates and tags within <open> tags.

Remove some extraneous whitespace in string_mixin.py.

Mostly by @gdooms, with tweaks.

- Stop parsing new templates if the template depth gets above
  MAX_DEPTH (40) or if we've already tried to parse over MAX_CYCLES
  (100,000) templates.
- Add two tests to ensure recursion works somewhat correctly.
- Fix parsing the string "{{" with the Python tokenizer; add a test.

- Newlines are now allowed in template param names.
- Changes also affect handling of arguments like {{{foo}}}.
- Update unit tests: remove some unnecessary ones, and add some to cover the changes.
- Update StringMixIn tests to actually work for some of the methods.
- Update copyright notices for the C extensions.

Some other changes, including removal of the 'string' import in the tokenizer.

Also, fix parsing of wikilinks in both tokenizers such that newlines
in any location within the title are an automatic failure.