Like MIME.decode_words_tokenized() , but the extracted atoms are
also remapped from their specified character encoding into UNICODE.
The result is thus identical to that of MIME.tokenize() , but
without RFC1522 escapes, and with all characters in UNICODE encoding.