Not an expert on this, but I would think about from a top-down approach rather than the other way round: So we start from a "fundamental" theory (FT), valid at all energy scales (high energy included) and we then derive the low energy effective field theory (LEFT), which is sufficient to describe physics at the low energies. There are two logical possibilities:
i)The LEFT is "of the same type" as the FT, ie the terms in the lagrangian for both of them are of the same type, or
ii)The LEFT is of "different type" than the FT, ie the lagrangian of the LEFT contains more terms that are needed to describe the effects of the high energy effects that have been integrated out.
If for a given low energy theory all the counter-terms are constrained by the symmetries to be of the same type as the ones we already have, then clearly we are in case i) and the theory is renormalizable. I do not see why the existence of a regulator would affect the argument. Maybe in one regularization scheme you can prove that the counter-terms are of the same type and in another one you cannot, so you should choose the scheme that works best if you are trying to prove renormalizability. However, this choice should not change the fact that the theory is renormalizable or not. It just makes it easy/hard to give a proof for it. Furthermore, I don't think it tells you anything about which scheme to use to actually make the calculations easiest.
Finally, regarding the statement "dimensional regularization is essential to the renormalization of QCD", I would interpret it as "dimensional regularization is essential to QCD, because it respects the gauge symmetry" and not as "dimensional regularization is essential to prove the renormalization of QCD".
Hope this helps!
This post imported from StackExchange Physics at 2014-03-31 16:07 (UCT), posted by SE-user Heterotic