I don't claim to be any type of expert here, but as I understand it, upsampling does not correct errors in the time domain like MQA does. Upsampling can have advantages for the filtering process but is not a panacea for reconstruction of 44.1KHz recorded digital audio. My subjective testing in listening to MQA encoded 44.1KHz audio vs straight PCM is MQA definitely sounds better. You could argue that a better DAC would rectify this for the PCM but the point of MQA is to enable this without requiring an expensive DAC with more sophisticated reconstruction filters.
Archimago is known, and freely admits, to being biased against MQA due to the lack of transparency and general business practices. The measurements shown are of audio "content" from 17.5kHz to 45kHz. MQA is not designed to encode this, as it's designed to encode music within the audible spectrum. So this test appears designed to make MQA look bad due to his personal bias against it. I wouldn't place too much importance on it.
There appears to be propaganda on both sides of the argument, as demonstrated above.