i think part of the problem is that nobody knows because sufficient experimentation has not been performed.
for example, here is the ib in a 4 cubic foot enclosure (not including inductance effects).
the native qts of the driver in that enclosure is 1.24, which creates the hump in the yellow plot. most folks would say that enclosure is too small because the qts is way too high.
but by simply adding a single band of parametric eq, 45hz, q=0.7, gain =-4.0db, the hump is gone and the shape is now that of a qts=0.8 (blue) which most folks would say is a little tight but satisfactory.
i do not know how far you can push this method. i have not seen it addressed either theoretically or empirically.
by making the enclosure too small, low end sensitivity is lost, but if that can be lived with, can simple eq "fix" the high qts or any enclosure that is too small?