That wouldn't be economic, even if ff sensors would get cheaper because of larger production volume. Unlike a computer cpus, afaik you cannot just software-fix faulty sensors on the silicon wafer, meaning smaller sensors = cheaper. And that's what you need if you retain any foothold in the high-volume entry-level market.
Quite true. However, in my opinion the term "entry-level market" no longer applies to the same demographic that it did in the previous decade. Digital photography is no longer the latest gadget - phones with pretty decent built-in cameras saw to that, as well as killing the P&S - and the people who now buy dedicated cameras buy them because it offers more than the imaging capabilities of their latest electronic gadget.
Also, as technology moves forward, consumers expect more capabilities at lower prices from electronic devices. In my opinion the current "Rebel" line just simply doesn't hack it anymore. However, the fact that they do still sell moderately well, just proves that people will buy anything they're sufficiently told to buy and that reality has quite hit them yet.
What I'm saying is that I agree with you regarding the high-volume entry-level market being important. I just kind of disagree on the weight of "high-volume" and what exactly defines the entry-level market. Personally I see the 6D as an entry-level camera now and for the next five years.
Another aspect with mirrored cameras is the size: Larger sensor = larger camera. What good is ff if you cannot tell the latest aps-c from ff up to iso 400? Last not least, ff is more difficult to handle due to the smaller depth of field. Enthusiasts may rave about creamy bokeh, but lots of people want infinite dof = smaller aperture = diffraction = less iq or at least no advantage to ff.
Yes, everyone wants something else. However, theoretically, if Canon dropped the "crop-frame" system and went with "full-frame" exclusively, then they would be in a better position to cater to more diverse needs.