The point is that if you are going to refute the large number of doctrines under the title "Christianity" then you ought to consider the best and strongest ones. ...But why should should I be primarily concerned with more sophisticated versions of Christian truth claims?
If you are going to refute Christian truth claims, then you ought to refute the strongest possible claims. The strongest possible claims are not constrained by Biblical literalism (because assuming that all Christians are Biblical literalists is an unfair assumption). So, the atheist ought to be primarily concerned with more sophisticated versions of Christianity.
I would accept this stance if I were concerned (hypothetically speaking) with refuting "evolution". Refuting straw man versions of evolution, while a popular pastime among creationists, is not a fair way to evaluate the evolutionary sciencies (the various scientific disciplines and their conclusions that fall under the umbrella of "evolution"). To criticize "evolution", one must, in a similar manner to what jeff.maynes suggests, criticize only its strongest arguments (and especially not consider a refutation of an obsolete argument to undermine the modern version of the science.)
But the rub is that "evolution" has a distinct canon: the scientific papers accepted as true by the vast majority of professional self-described evolutionary scientists. This canon, moreover, is highly internally consistent as well as highly consistent with a well-defined metaphysical epistemology (the scientific method). In short, there's something specific, well-defined and there to criticize.
In fact, it's a misnomer to assert that one should only rebut the "strongest" arguments for evolution. Rather, one should only rebut the canonical arguments for evolution. They do happen to be the arguments considered strongest by the canonical authorities (the professional scientists), but it is their inclusion in the canon that gives them privilege, not their strength.
Another rub is that the "umbrella" of evolution does not in any way protect weak arguments actually made or exempt them from any criticism. If, for instance, a critic of evolution in general were to correctly rebut a false argument in evolution (e.g. adaptationism) actually made by someone naive, then it would not be fair to then criticize the critic for rebutting a non-canonical argument. Whether or not it's canonical, the position is false and deserves criticism, and criticism of a part is no less deserved or valid coming from a critic of the whole.
The problem with jeff.maynes' remarks about "Christianity" become clear. There is no single canon that deserves the privilege of being called the Christian canon. There are, rather, a thousand different canons, all held to a differing degree by self-described Christians; every Christian has his own individual interpretation for what it means to be "Christian". There is no privileged canon of Christianity as a whole to which one can legitimately claim that criticism should be restricted.
Even if there were a canon, it would not be legitimate to use that canon to shield weak arguments actually made. Again, if someone advocates adaptationism, they should be corrected; likewise if someone advocates Biblical literalism, Aquinas' five proofs, Anselm's or Godel's ontological arguments or Pascal's wager, they should be criticized—even if these arguments have been rejected by some theists, even if there were actually a Christian-as-a-whole canon that rejected these arguments. And it is a fact that millions of people, if not hundreds of millions, do actually believe and put forth these arguments.
One can attack and defend evolution because evolution is something specific. One can neither attack nor defend Christianity, however, because there is no one, well-defined thing, however abstract, that is the Christianity. All one can do is discuss the specific arguments and truth claims that actually arise in real conversations. And one cannot use even a canon to shield non-canonical claims from criticism; to use a non-canon to shield such claims is doubly fallacious.