The Healy-Gilchrist situations shows why Boucher got such a long run. It's hard to drop a keeper. If it ain't broke, don't fix it. You need a lot of evidence to suggest you're getting an upgrade in runs without losing too much behind the stumps. Pakistan chose Kamran Akmal who averaged about 2-3 runs per innings more than their previous keepers, but cost them multiple tests with his ineptitude with the gloves.
South Africa could have replaced him with de Villiers who was likely to score more runs than him, which would have freed up a spot for another batsman. Who would that have been, and would the (new batsman vs Boucher) gain be more than the (de Villiers' runs as keeper vs de Villiers' runs not as keeper) potential loss? Not all batsmen who take the gloves see their runs diminish - Flower didn't. But generally (Walcott, Stewart, Sangakkara) the evidence is that they will. And so if you're turning a great middle-order bat into a 40-average keeper so you can play say a mid-30s player like Petersen or Duminy in preference to Boucher, are you gaining anything?
Was going to say that Boucher always had Pollock behind him at 8 so that insulated him a bit, but just checked and in 2008-09 they beat Australia 2-1 with an 8-11 of Harris, Morkel, Steyn and Ntini.... and of course in the second test that lower order took them from 140-6 to 459...