Microlenses are not the enemy, but rather the pixel stack height. A pixel on an MF sensor today is about as high as wide and the photo diode is at the bottom of the well, meaning that it can't handle low angles of incoming light - that is the problem. Microlenses actually help that somewhat, small pixels without micro lenses would probably perform even worse (although we would probably not get the ripple).
The older microlens-free sensors also have larger pixels, which reduces the height/width ratio, which is the key why they work better. The lack of microlenses is not all good though, it means that there is more "empty space" between pixels where no light is captured, ie you get more aliasing than when the light is steered from the edges of the pixel into the photo diode. You also get a lower ISO but for the wrong reason (light thrown away rather than a bigger light capture capacity).
The main reason microlenses has such bad reputation is the early generation ones used in the Kodak sensor of the P30+ and others. The microlenses on modern sensors are much better. But better still would of course be to place the photo diode on the surface of the pixel and have it cover the full pixel area, but sensor manufacturing technology is not there yet, at least not for MF-sized sensors.
Concerning retrofocus lenses it's revealing to look at Phase One's 645DF wide angle lenses without C1 lens correction algorithms applied (use rawtherapee for example), there's lots of chromatic aberrations. With tech lenses there is no such lens corrections, there cannot be since there is no information in the exif data like shift and tilt settings or even aperture and focal length that would be required for that to work. Tech wide lenses thus uncorrected exceed the performance of 645DF wide lenses even when they are corrected (concerning sharpness), and that's impressive.
The reason this is possible is not some magic superiority in the manufacturing process, but a matter of optical design formula. These lenses have less maximum aperture and are less retrofocus (or even symmetrical), then you can make sharper wide angle lenses, with less barrell distortion too. The drawbacks of this optical design are smaller max aperture, more vignetting (requiring center filters) and being tough on current sensors due to low angle of incoming light off center.
As tech cams won't provide exif data of lens and movements settings anytime soon (that would require collaboration between multiple tech camera makers, an electronic shutter maker and digital back makers, ie it won't happen), software correction of the optics is not feasible, so simply having superior optics will still be required well into the future.
If due to sensor design tech lenses much be more retrofocus in the future (eg if CMOS sensors take over and there's no solution to the crosstalk problem) you'll either need extremely complex optical formulas (making the current Rodenstock 32HR seem light, trivial and cheap
) or more compromised lens types which would give less good sharpness than the 645 mirror cameras.
Personally I believe the traditional "large format" lens designs have a place in photography and I hope that CMOS sensor technology improve on this aspect before it's too late. The SK28 was discontinued not because it's a bad lens, but because the modern backs don't support that lens design. Much more lenses in the current tech cam wide angle range would need to be discontinued if CMOS with current performance becomes the new standard.
Thanks for test file offer, I have already got a bunch of test files which I have not had time to sort and look at yet so I'm not yet sure I need any more than I got. I just need more time
, I'll get back to you if I need more files later.