Op-amps are far from perfect, even though the textbooks would lead you to believe it. However, like most things, the answer is "it depends". It depends on how the power for the op-amp is treated and what op-amp you are using. I have done a lot of experimenting with op-amps in the past and learned a lot of things that are not in the textbooks. If you choose the right one and treat it properly in the design, it can be just as good as a discrete circuit and sometimes better because the layout of most discrete circuits does not allow for optimum power delivery. The compact nature of chips allows for optimum power delivery.
Sure, I have also experimented extensively with OpAmps. The only one I currently admire as much as a discrete circuit is the newer MUSES01 by JRC. But I use it in my phonostage.
Even the reduced out-of-band noise of Sigma/Delta DACs calls for a very high slew rate OpAmp, but those never delivered what I expected. My opinion, of course.
I never claimed that S/PDIF is the best method for DAC input, but again the answer is "it depends". If the circuit is well-designed, a good coax cable is used and the right receiver chip is used, it can be every bit as good as the best internal USB interface. So close that no one can pick it out in A/B comparison. I have done this.
I was referring to the Berkeley claims. Of course, you are correct that USB can be as good as S/PDIF (or vice versa) if done correct. That is also my experience. But USB offers a much higher resolution that is desirable, at least for me.