What design philosophy favors rising distortion at higher frequencies?
I don't know, I was just trying to respond to the question about if Marantz picked that filter because it is the default for that chip. So the fact is, Marantz tech support confirmed it was intentional when I asked them. I actually posted their response a couple times before and iirc their explanation for that choice wasn't clear to me. May be you should ask them too and see if you can get a more detailed explanation.
I also quoted Gene's (audioholic.com) understand of Marantz's rationale (he said he got it from Marantz), but all I did at the time was quoting what Marantz told me, and what they apparently told Gene. Whether any of those would make sense to anyone would be up to the individual. For me it does not make much sense, but that's just me and I have tried different filters using a couple of my DACs and I could not tell a difference anyway.
Also, my understanding is that the issue may be more to do with frequency response, not so much "rising distortion at high frequencies" except there may be issues related to the topic of "pre ringing, post ringing..", that I am not familiar with.
Other than that the rising distortions you referred to would be in the range above 20,000 Hz. You can see that Amir used 90 kHz bandwidth to show the "rising distortions" that increased to about 0.02% at about 10,000 kHz, if he had used 48 kHz you would not see much of the effect.
Using 90 kHz BW, the sampling frequency of the test signal would have to be much higher to make the effect of the filter disappears, and Amir noted "same but at 192 kHz sampling removes the effect of DAC filter". I would think that even at the lower 48 kHz sampling frequency, the distortions within the 20,000 Hz range would be much lower than what the graph shows because the graph shows the
total harmonic distortions. The 2nd harmonic of 10 kHz is already at 20,000 Hz so the 3rd would be in the ultrasonic range already.