I think these are good observations. As you say, the KDE developers not only wanted their applications and accessories to have the same look and feel, but they also wanted to provide a framework for developing applications and the common functionality that one expects from a desktop environment.
In turn, this fed the demand for KDE, as opposed to just running a bunch of applications and a window manager, because one could be sure that dragging something from one application to another, for example, would probably invoke a common operation (like dragging pictures from Digikam to something offering a file management interface invokes a dialogue about moving, copying or linking the pictures as files, which they are even as far as Digikam is concerned) rather than the drag operation not doing anything or doing something weird.
Cross-desktop integration has improved, but there's still the issue that various applications (including some of the mainstays of GNOME back when KDE seemed to offer its own coherent set of applications and GNOME clearly wasn't able to offer something comparable at the same level of integration) don't take proper advantage of core desktop functionality. I did run OpenOffice with KDE dialogues for a while, but it was a flawed experience.
It's easy to make the observation that GNOME in particular has backed even further away from the original KDE vision of a coherent suite of applications and desktop components, but if that is the case then what does that leave for the project's developers to work with? The start menu, some dialogues and whatever happens when the user presses Alt-Tab? The stark choice at that point is to either take a back seat and be like XFCE or to try and make that functionality more prominent somehow, but there has to be a good reason for doing the latter.