The rationale as I see it is that you can purchase a single good scaling and deinterlacing solution and use it for all of your sources - which is why standalone video processors came into being in the first place. What's happening is we have started seeing those processors get shoe-horned into surround processors. The difference between average and exceptional in this field is potentially huge (especially for deinterlacing, although scaling and other features such as noise reduction also contribute).

Personally, I go back and forth a bit these days. On the one hand, a good standalone video processor like the DVDO Edge that was able to get instructions from a receiver or processor (to automate the video switching) would be a great solution that would also fit with the traditional "separates" model (replace one piece of the system without paying to replace others). On the other hand, we do end up with a lot of components, remotes, and cabling to manage in that scenario that could be simplified by integrating the video processing into the receiver or processor. As always, there's no one perfect answer...