Neuro, I find that my views often coincide with yours, but not on this point. I think the photo-video convergence isn't something artifically forced into existence by canon; rather, they had a good read on the direction the industry was moving in and got a great jump on it.
In terms of the ideal use of r+d resources, I'd like to propose an analogy to another market that delivers similarly complex, multifunction machines- the auto industry. If we think of cars twenty years ago, they were relatively spartan devices that revolved pretty much around an engine, chassis, suspension, and simple cabin comforts like air conditioning. Since then, based on market analysis and the desire to produce unique product, manufacturers have introduced dashboard computers, power seats, entertainment systems, panoramic sunroofs, bluetooth voice recognition software ... none of which really contributes at all to the process of driving a car. Yet at the same time, they've also improved by leaps and bounds in engine, suspension, and chassis technology. Would a BMW 3-series be getting 40 mpg if BMW weren't wasting engineers on fixing the horrid iDrive software?
While opportunity cost always exists, I don't think its always so simple. As studio observes, without video implementation the 5D mark ii may very well not have brought in as much money for canon to reinvest in new technology. We can't know for sure. I am firmly of the belief that canon can improve video quality as well as stills performance at the same time. Besides, better DR and lower noise improves video image quality just like it improves photo IQ ... so again, its not a mutually exclusive use of resources on canon's part.