but one sort per tab and none per list is arguably better than
           O(n + n**2) per tab and O(n**2) per list.
 OMG, someone shoot me.
 ?
 I can't believe I just used the big goose-egg to explain why my
           way is probably best in the long run.