Yeah, but that was a purely resistive load. And there is no swing.
Here's what happens when the O2 is subjected to other loads. Also the second part of his measurements:
http://scientistsaudio.blogspot.com/2012/06/leckerton-uha-6smkii-part-2.html
And you are right that it doesn't follow that any 600 Ohm headphone would be dominated by resistance, but doesn't that practically mean that every "frequency response graph" made with purely resistive loads is meaningless since a real headphone would have both inductance and capacitance playing into its impedance swing?
I would call those IEMs a "hard load" because of the impedance swing due to the crossover. Another hard load would be the other extreme where
Even with very low output impedance, the O2 is doing a worse job than the Leckerton.
Also in the case of the DT880, which has an impedance swing up at the higher extremes, perhaps the iQube would be able to equalize that and make the resulting output frequency response more linear. Otherwise it should be rated worse than both the Leckerton and O2, but that's not the case.
So again, I'll say this: the frequency response graph of a headphone amplifier is not the only indicator of its performance.