The problem though Steven, is that reality contradicts this, and faster lenses down to f/1.4 etc do actually result in brighter images....
It's a simple fact of physics... a lens of any sort has an associated (max) aperture, and that means it has a max transmission capability. But one thing to note is that aperture is not a hard limit, it's essentially a "percentage rating." I.e. if the primary lens projects more light at a wider aperture the micro lens will also transmit more light, just not all of it if it's aperture is smaller/more restrictive.
The difference between the focus screen and the sensor's micro lens array is that the focus screen is a single lens (essentially a fresnel lens). Because it is a single lens it has it's own aperture which limits both the minimum DOF *and* max light transmission. This is also true of the AF module which typically has an aperture of approximately f/7 in terms of light transmission/DOF (per lens/ AF area).
But the micro lens array is not functioning "singularly." Even though each has it's own aperture which limits light transmission per pixel, it does not limit DOF *per image*... the DOF per image is a matter of what falls where, not how much (similarly, SS does not affect DOF).
The image does indeed record brighter at wider apertures, but that is largely due to unreported amplification which is required in order for the exposures to follow the convention of film... I understand the f-stop and t-stop (ratio) are relative to the FL which is only (kind of) accurate at infinity focus where FL is determined/measured. But even if that isn't accounted for the relative light losses measured, and the increased gain measured means that the appropriate amount of light is not reaching the sensor (i.e. you couldn't use the loss to determine an accurate t-stop rating, but it's still real). This is separate from t-stop/vignetting and only applies to digital.
I do know that, in general, there have been improvements made in micro lens arrays/fill factor... so maybe no modern camera sensor (pixel) is limited to f/4 as the D200/7D are in those tests. But I doubt the issue has been entirely corrected because it seems like at the same time they increase efficiency they also keep increasing MP's... maybe one is only allowing the other w/o making things actually worse?
None of this affects any of the other aspects of the fast prime on digital (construction/sharpness/DOF/etc)... but it is kind of messed up IMO. Probably best to put it in the category of "I can't change it, so there's no point in worrying about it."
Edit: I suppose my initial statement was a bit misleading/incorrect... I did say "the sensor doesn't actually get more light."
OK, entirely incorrect....