Have you considered simply adding up the pp for the different individual combo sets for tag4 (as in, instead of counting it as one map with lost of insane jumps, it is four maps (one for each color) with lots of time to move your cursor).
That would require per-hitobject data. And it again would be special treatment for specific maps. There simply is no way to unify touch and cursor-based input in one difficulty system. What's hard for one is easy for the other and vice versa.