That isn't an argument against MWI/Everett, it's an argument against thinking about quantum computation as "computing 2^n values of f(x) in parallel in different worlds". (On the grounds that what licences us to consider two Everett branches as separate "worlds" is that decoherence makes their interactions negligible, while what happens in quantum computation is all coherent and preventing decoherence is a central engineering problem. [EDITED to add: And that the operation of a "cluster state quantum computer" is not helpfully thought of in terms of doing lots of computations in parallel. I don't know enough about the cluster state model of quantum computation to have any idea how strong that argument is.])