I've used to think about the mismatch this way:
If you have IR intake, it's beneficial to have "trumpets" in the end of the duct, because air likes to go in from the sides of the duct rather than from the center.
Big intake to head mismatch, where head entry is smaller than intake exit, is a bit similar situation as I see it. At the discontinuity air wants to enter from the sides more than it otherwise would, evening out velocity profile and possibly pulling in fuel that has been wetting the walls and collecting to the step. Small rounding in the head entry could be beneficial to make it resemble the trumpet a little from flow point of view.
It would be similar phenomenon that exists in more continuous way in the funnel shaped duct. It would be interesting to find quantitative info about size of the step (or angle of funnel) impact to velocity profile.
I'd bet that having a small rounded step "in the perceived wrong direction" meaning head entry is smaller than intake exit, will generally make more power than having it vice versa, which tends to be more typical approach (head entry slightly bigger than intake exit). But what do I know, I haven't tested it, and it's easy to get anything wrong without testing in such a complex system.