The choice of whether to include a work is based on whether its license is free. The definition of "free" is based, ultimately, on whether it benefits free software or not. You're trying to bypass the process that determines that, by handwaving wildly and saying "but anyway, who cares, it would benefit free software to make an exception for this thing and that thing". Sorry, but you're just not presenting any arguments that I think are worth spending further time debating. If anyone else thinks this has substance worth discussion, they're free to jump in, of course.
I completely disagree with your definition of free. Moreover I have never seen such a definition (not in the FSF, nor in the open source movement, nor even in the DFSG or the Debian license tests). A license is free if you can practically use your freedom, not if it is ideal. You completely mistake the fact of promoting a license and accepting one. Finally I do not claim that we must make exceptions, I claim we must interpret free less strictly than Debian legal (much on the same way as the open source movement and the FSF): it is astonishing that licenses that "does not follow the DFSG" does follow the law of the open source movement which are exactly the same ones!
It's unfortunate that you place so little value on free documentation, but that's your choice. I hope the free software community at large ultimately disagrees with you.
I place value on free documentation but not on your definition of "free". GFDL can be modified, with the inconvenience of being obliged to include invariant section (which are non-technical).
Finally, both the FSF and the much bigger open source movement agree with me (more modestly, I should say that I agree with them), not with you. With these absurdly strict policies, Debian eventually does not agree with itself: it's own logos cannot be modified! which show that these policies were not what Debian want at its creation.
Olive