My point was meant to be that if that was their intention or even if they were going to accept that (not worrying about sales/profit form the 1DX) they would have designed one sensor for the 1DX and 5D3 and not spent all the money on R&D and setting up the fab. This view of course is predicated on the rumours that the 5D3 will remain in the 21-22 MP range... if it ends up being higher then of course they might not have been able to get high enough fps from the 1DX.
I agree - a 21 to 21MP sensor for the 5DIII doesn't really make sense economically, since it would have no real IQ advantages over the 18MP sensor in the 1DX.
I believe that with the 1DsIII and 5DII, Canon probably made a very profitable move by allowing the 5DII to cannibalise 1DsIII sales. They retained the 1DsIII as a flagship for the small niche market that really needed its features, while the 5DII has been selling like hotcakes. I don't think it would hurt Canon to make the 5DIII be very close in feature set to the 1DX - provided the price is right for it to sell in high volumes, but there is enough differentiation to ensure that high end users do not totally abandon the 1DX. Assuming that the sales volumes of 5D series cameras are possibly an order of magnitude or more higher than 1 series cameras, Canon stands to recoup development costs faster on a 5D series camera, even at a lower price. If a lot of components are shared, that even allows the 5D series to bear a lot of the development costs of the 1D series. - This would be a bit more like the auto manufacturing approach of building around a platform - where for instance the Audi TT and VW Golf share the same platform, but target very different markets and sell at very different prices. The TT, on its own, could not justify the platform development costs however.
... Just my speculation.