multiple ?w behavior in hybrid mask mode

Currently if you use more than one ?w in a mask the same word is repeated.

Wordlist = (One, Two, Three)

mask=?w?w?w

will yield

OneOneOne
TwoTwoTwo
ThreeThreeThree

I would have expected it to yield

OneOneOne
OneOneTwo
OneOneThree
OneTwoOne
...
ThreeThreeTwo
ThreeThreeThree

While outside scripts (example: https://github.com/LordDarkHelmet/WordListHelperScripts) can be used to create permutations for users who have used a combination of multiple words in random order, it is an issue for users who use multiple words where the words are not next to each other.

--mask=?w?d?d?d?d?w

It could be a nice feature to allow several different word lists in a single mask. This will keep the existing functionality of multiple ?w and allow for greater control.

For example ?{w1} references word list 1 and ?{w2} references word list 2, ... ?{wN} references word list N

--wordlist1=wordlist1.lst
--wordlist2=wordlist2.lst
--wordlist3=wordlist3.lst
--mask=?{w1}?{w2}?{w3}

In the above case the user has a word list for each "slot". wordlsit1 could contain names of people, wordlist2 could contain names of pets, wordlist3 could contain street names.

--wordlist1=wordlist1.lst
--wordlist2=wordlist1.lst
--wordlist3=wordlist1.lst
--mask=?{w1}?{w2}?{w3}

In the above case the user will get every combination off of a single list because each of the ?{wN}'s are assigned to the same word list (wordlist1.lst), but iterating independently.

It would also allow for

--mask=?{w1}?d?d?d?d?{w2}

where the first word is different from the second word.

Start by locating the mask and wordlist handling code in John the Ripper; the issue does not name specific files or tests. Define completion as independently iterating multiple word-list slots, while preserving existing repeated ?w behavior, and verify the documented ?{w1}, ?{w2}, and mixed mask examples.