Use utf8 throughout...

[lyda]

One of the words from the misspellings list that is not included is gardai -> gardaí due to the fact that utf8 is not supported.

1. Change the code such that it is.
2. Add a test case including "gardaí".
3. Revisit commit 475fe970cde7034102f240e432b137d65639635a and _WORD_REGEX. Change it to do a findall and have it list the chars it allows in words rather than the chars that are not in words. Actually it could even dynamically generate that list from the wordlist it is using.