// returns all the pairs, but best matches first
// second part of pair is similarity score (0-100)
static LPair<PairS, Int> mostRelativelySimilarLevenICPairs(LS words) {
  new Map<PairS, Int> map;
  for (LS l : allSubsetsOfSize_iterator(2, words))
    map.put(listToPair(l), levenSimilarityIntIC(first(l), second(l)));
  ret pairsSortedByBDesc(mapToPairs(map));
}