This new Canon is actually a 200-560 f:4-5.6, in two stages. Would it actually have been easier to just make such a zoom, or is this solution better what comes to IQ? In a longer zoom the inner movements are longer and more difficult to keep tight tolerances, but with this built-in extender there are (probably) more lenses and also a sideways movement with the extender part.
I suspect that 200-400 f:4 with built in extender also sounds more professional than an aperture loosing 200-560 f: 4-5.6, even though the later would be more convenient in actual use. Professionals are a conservative lot. When they came up with the idea of a built in extender maybe nobody said out loud that it really is not a great idea compared to a continuous zoom, as it was designed form a clean slate anyway.
As a teleconverter essentially "pre-crops" the image before the image sensor (enlarging lense faults), what is the opinion on using e.g. a 7D on a 200-400mm vs using e.g. a 5Dmk3 with a 200-400mm with teleconverter?
I would assume that a key question is if it is the lense or the image sensor that is the main limiting factor. If sensor noise is the main issue, one would perhaps want to use the larger, better 5Dmk3 with teleconverter. If sensor sampling grid is the problem, one might want to use whatever sensor has the highest number of megapixels (likely a larger sensor one, but 7D vs 5Dmk3 is a small difference). If lense aberations (large aperture) or diffraction (small aperture) is the main issue, then one perhaps might as well use a crop sensor. A large, dense sensel camera (like the D800) might offer a sensible compromise between both.
If you have got the 200-400mm IS TC, then such questions might not matter. But if you have a 70-200 they might.