This is more relevant than your other citation but it still doesn't support your claim. This paper describes evidence that mutation rates are lower in critical genes, reducing the rate of deleterious mutations. It's quite interesting, but not really all that surprising -- there is broader evidence for evolution of mutation rates, and the structure of the genetic code itself is well known to make deleterious mutations less likely.
What the paper doesn't do is make your claim, which is that some mutations are programmed because they are potentially useful and that deleterious mutations are not programmed. Rather, they find that rates of all mutations, good and bad, are higher in some regions and lower in others. In fact, it specifically rejects the possibility that mutation rates are higher where new variation would be useful, noting that antigen-producing genes (where variation is very useful for evading host immune systems) do not have elevated mutation rates. Instead, what seems to be happening is that there is strong selection against having mutations in certain genes, since they're almost always deleterious, and that mechanisms have therefore evolved to keep mutation rates unusually low there.