Regex for separating lemmas may fail in some edge cases

[scottkleinman]

This is a leftover from issue #1013. In scrubber.py, line 169, the regex pattern is supposed to detect whitespace, unicode punctuation, or the end of the string. See the comment in the preceding commit. However, this commit did not supply a pattern for detecting Unicode punctuation. As yet, I have failed to find a concise way to do this, so I have supplied \W (non-word character). This works pretty well, but it may fail in some edge cases. This needs further testing.