AFAIK you actually are more inclined to make these splits in higher counts. For example 3,3 v 8 is a basic strategy hit but becomes a split at +5. So yes, to the best of my knowledge you do continue to split 22 v 7 and 33 v 7 at higher counts because if you do happen to catch a seven, eight or nine then you have an even more profitable double on account of the high count. Presumably there is some extremely high true count where hitting becomes the more profitable (or rather less unprofitable, to use a double negative) option but it's truly academic because it's probably so high that it would almost never occur and pair split indexes in general are less important because pairs are pretty uncommon in the first place.
