Similar to MIME.tokenize, but lables the contents, by making
arrays with two elements; the first a label, and the second
the value that MIME.tokenize would have put there, except
for that comments are kept.
The following lables exist:
| "encoded-word" | Word encoded according to =?...
|
| "special" | Special character.
|
| "word" | Word.
|
| "domain-literal" | Domain literal.
|
| "comment" | Comment.
|
|
MIME.tokenize()