There is no such evidence. There are a million ways to get 0.004% THD+N so what you say can't possibly be correct.
Look, here is the problem from company's website:
View attachment 146353
People read this and assume not only is it transparent to CD but will reach even higher levels of objective performance. Yet it has second harmonic distortion that exceeds CD's noise floor:
View attachment 146354
By contrast, here is the performance of Apple's $9 dongle:
Its second harmonic is down to -100 dB, easily clearing the noise floor of CD. Indeed you can see that in the SINAD of 99 which is a whopping 12 dB better than Bluesound NODE. It also has almost no harmonic distortion past second unlike NODE.
So please dispense with making excuses for manufacturers. They don't do a better job because people like you do casual listening tests and rave about what they hear. That gives them no motivation to do better.
And no, doing better doesn't mean higher cost as the Apple dongle shows. And countless other budget DACs we have seen with much improved performance with zero impact to their retail cost.
You can wash your dishes once and do a good job and they come out clean. Or don't do a good job and with same effort, produce still dirty dished.