Words method uses punctuation emojis as a separator

[eabrakov]

For the case string has some unicode symbols the src/.internal/unicodeWords.ts counts all punctuation symbols as separator, even if it followed by \ufe0f for example, which could be used as an indication of emoji, but not punctuation

So

words('abc!def') // ['abc', 'def'] as expected
words('abc\u203c\ufe0fdefdef') // actual ['abc', 'def'], expected ['abc', '\u203c\ufe0f', 'def'], same as other emojis

Please let me know if you have no concerns about expected result, so I could create a PR with the fix