EDIT: GODFORSAKEN MOBILE AUTOCORRECT, sorry for the eyesore post folks.
Of course its measurable. People simply are looking in the wrong place. Ive been saying for a while now (one of the aspects you got correct was cup size and shape, which directly relates to pinna activation, of which there is no coincidence something like the HD800 excells in along with its idiotic treble rape that also suits slowly defending aged folk). The other aspects are related to recording quality (good clean mics), recording type (binural or not etc..), recording setting (massive difference when you listen to a recording done in a church hall, as opposed to a treated foam walled studio room). The rest has basically to do with post processing effects like channel panning, and of course the obvious echo and reverb and other DSP effects that all play on frequency response as you alluded to somewhat with a preference curve which of course is almost exclusively observed from a FR aspect.
Aside from these things that can be broken down into a bit more nuance if you really wanted to... There is no "soundstage" as a singular metric based solely on hardware design. Such a notion is ridiculous for two reasons. The first, businesses couldn't wait if they could legally and scientifically have another quantifiable metric to include in spec sheets. The second and more important reason is I would be able to take a mega soundstagey headphone and play mono recordings all day and still it must produce a soundstage effect if soundstage was truly a mostly hardware phenomenon.
You would also have such metrics provided by driver designers (and acedemic guidelines of what is strictly required in design to produce hardware exclusive soundstage-ability levels within your hardware design).
I dont trust a single headphone company does basic product validation to the level of someone like Amir does (they prossibly cover legal things like safety and such at best). I have more belief companies that contract OEMs for parts like Apple does with their IEMs for example do far more testing of what they're sent than any audiophile company simply in virtue of the stakes they play with concerining the level of investments they make requires such discretion.
And when I don't see them trying to chase this soundstage ordeal in mostly hardware, i feel safe with my notion of what soundstage is. To me soundstage needs to get measured at the production/software level, or if there isnt access to such, then it needs to be done like Harman research (preference exploration to see if there is any pattern that can be exploited and worked on specifically in-hardware). But with manufacturers providing incremental changes for each test phase to see if we're heading in the right direction.
You recall when i said one of the flaws in assuming soundstage isnt primarily at the hardware level? In software or at the recording stage of production, i can crank the Echo knob, or I can decide to record in a church. And even if i played the recording in mono, i would still get more spatial effect than if i tried to chase soundstage by simply trying to change the hardware of the headphone.
The reason i made my remark was because I thought it was a bit funny to see amir talking precise percentages of soundstage-ability. I know the subjective portion is just someone trying to use words people can understand rather than seeking out audio engineers to explain to them what 6Khz-30kHz frequencies fo with respect to soundstage. So i actually dont like when people give him a hard time sometimes with the subjective portion. Its just entertaining to see how someone with his technical vocabulary behaves when he forces himself to use the terms in a style mostly used be self proffessed expert audiophiles who also think "measurements cant tell me if ill enjoy the music" (as if thats what all measurements do or something).