Perhaps you can show measured the frequency response for the X3 and we can compare?
If there is a mid-bass hump on the iP5, it should show up in a frequency response curve. The only way I can see that it would otherwise show up is if the iP5 had a high impedance on the headphone out and the Atrio MG7 had an incorrect damping factor. But the iP5 has an output impedance of 4.5 ohms (a little higher than I'd prefer) and the Atrios are rated at 32 ohms - so damping shouldn't come into it.
If the iP5 was underpowered - this could also have an effect - but given that the Atrios have a sensitivity of 112dB 30 Hz/Mw - and the iP5 will be able to drive them to deafening levels well under the output full volume, then this is negated also.
So no issues with power or impedance mismatch on the IEM's he was talking about. So where does the mid-bass hump come into it? Simple answer - it can't come from the iPhone.
I understand what you mean and yes i've thought about it before and i really couldnt make sense of it. If you check out clieos review you can see his measurements. In fact some time back there was a discussion where he mentioned that there was no such midbass bump but others claimed very strongly that they did.
I really do not have any idea how it works i'll be honest. But i thought about it and the only conclusion i could come up with was that the measured frequency response curve of the source and amp do not say much about its signature. Perhaps it might be that measurements are made with sine wave sweeps and not does not simulate the harmonics of actual music? I really have no clue hehe its all speculation but i know for a fact that the curves do not show its sound signature