I am so happy not to have based my purchase based on ASR review (I use to spend hours reading everything when looking for something, and ASR or AH reviews are awesome compared to subjective reviews). I based my purchase upon features, reliability and price. And what a surprise when I finally got my Yamaha to replace my Denon old very high end AVC-A1D : my speakers finally sound amazing! Maybe it’s a matter of sound taste, speaker pairing I don’t know, but it’s not a 0.1% that defines the sound quality when I am listening music, my ears aren’t good enough. I wonder if instead of distortion, it’s not something related to fast changes in volume that should be tested, what we call detailed and dynamic sound. Measuring the ability to reproduce a sound without distortion in a sweep test is something, being able to being nimble is something else. It’s like looking is a car is able to maintain the perfect speed in a linear relationship between throttle and real speed while the real use case and performance is how a car accelerate and brake.
Even if the Denon AVC-A1D is in top shape, based on the audio specifications only it will lose, except if everything is kept in the analog domain, then it may have a chance.
The thing is, specs and measurements are not affected by expectation bias and Placebo effects that can only be taken out of the equation if the often mentioned (like a million times) tightly controlled double blind test protocols are followed. Even for comparing loudspeakers (that we all agree most can tell the differences), tightly controlled blind test results had shown that sighted tests were unreliable. Even the color of the speaker grille made a difference in "sound quality".
The Dishonesty of Sighted Listening Tests vs Double Blind - YouTube
Your point about the "fast changes", "dynamic" are obviously part of the equation, but the associated factors/specs such as slew rate and damping factor have been debunked many times by various experts, not because they are not relevant, but because those are no longer the issues, or limitations especially when you are comparing the mid range AVRs that pretty use the same bunch (or very comparable) of ICs used in the audios signal chain while the design of those signal chain are highly standardized, not rocket science involved there.
So you are right that the difference between the 0.05 or 0.005% THD+N are not audible to the non golden ears group, but the same goes for the damping factor and slew rate, in fact to the point most test benches including AH and ASR would only measure them occasionally. Audiophiles do continue to cite things that are not measured as reasons for what they claimed they "heard", the last popular one seems to be TIM distortions, transient intermodulation, that is also one related to your last point, but when was the last time AH or any test bench even bothered to mention it, let alone measuring... Again, yes AH and ASR can, probably should measure a lot more to try and find out what's causing the audible differences, but my educated guess is that they wouldn't bother because deep down they know the "truth".
I also would not make my choice of my next AVR or AVP based on AH or ASR's test results only, but on "all else being equal, or highly comparable" basis, all else would include features, perceived reliability, ergonomics, even the look, then yes I would base my final decision on specifications and bench test results.