So, i've been bored at work today, reading stuff about lenses (like every other day, really), and found some more interesting information on the problems on the angle of incidence.
I've come across
this article. It's a fairly long-winded history of wide-angle lenses from Zeiss. But read page 12 in particular (you can just skip straight there if you know the difference between a Distagon and Biogon).
The long and short of it is this:
Take a regular wide-angle lens for an SLR, FF or APS-C, the light waves are going to hit the sensor a lot closer to orthogonal (that's a right angle, fyi).
Reduce this down into a mirrorless design, the lenses will be smaller, because there's no mirror in the way, right?
Well, maybe. If you want a smaller lens to put on your thinner FF mirrorless body, it's not as easy as just reusing the same sensor from your 5D. I already knew that leica had some fancy micro-lens arrays going on in front of their sensors. But according to this article, you also need a much much thinner IR Filter (and AA filter, and bayer array, and protective glass) in front of your sensor. That's probably why leica 'forgot' to put in the IR filter on their first M8. That's also probably why Leica has no AA filter. It's not for better IQ like it is in the D800E, it's because there can't be one. Well, OK, there can, but you're going to have to pay for it. (maybe this also explains that $3000 Sony FF P&S? Any cheaper and it'd be huge).
Leave in the regular-sized IR filter, bayer array, protective glass from your 5D, put it into a thin mirrorless body, with a thin pancake wide-angle lens? Your corners will not only look horrid, but you introduce field-curvature too.
So, at the end of the day, what can you expect?
You'll get an FF mirrorless eventually, but it's either going to be the
size of the 5D3 (and the lenses aren't going to be any smaller either), or it'll be the
cost of the 5D3, or it'll have so bad corner performance that it may as well be APS-C anyway.