Bug(medcat): CU-869ag0tqj Fix regex tokenizer dashes

[mart-r]

The regex based tokenizer would treat all words that had some punctuation in them as skip words and would skip them.
And because of that, these tokens would be ignored since the tagger would tag them as skip words because they contained punctuation.

This PR fixes that by separating the punctuation in the regex based tokenizer example.

This should fix #128.

EDIT:
To be a bit more explicit regarding the regex tokenizer:

• Before it would tokenize know-how into know and -how
• Now it will tokenize know-how into know, -, and how

[tomolopolis]

Task linked: CU-869ag0tqj Fix issue with dashes and regex-tokenizer

[alhendrickson]

lgtm, now it's much simpler with your latest commit