...it is easy to see that a new 10MP chip could equal or even exceed the 5D's older 12MP design since that is even less of a difference at 18% or so...
I take your point and if this were to happen I might find myself in a similar position to yours when the 5D was released, ie. struggling to justify the superior resolving power of the much more expensive 1Ds2 .
I suppose the outcome will depend largely on the noise characteristics of those slightly smaller pixels. If Canon pull some more rabbits out of the hat and succeed in making the 400D pixels even less noisy than the 20D pixels, perhaps through greater quantum efficiency, then we can probably look forward to some protracted arguments comparing the 400D image quality with the 5D, especially considering the 400D will be accessing the sweet spot of the image circle.
However, we should bear in mind that, as pixel density increases, the sensor struggles to record the higher resolution from the lens (which usually exists) because of falling contrast or lower MTF of such higher resolution, therefore the increase in recorded image resolution is not proportional to the increase in sensor pixel density. There's a 'law of diminishing returns' at work.
Whatever the outcome, 5D owners can remain smugly content that their wide angle lenses are truly wide angle .