".... But there must be some point when tread design has a more significant effect on fuel economy than just mass...."
I'm sure there is a point, but it must be WAY, WAY out there, because the stiffness of the belt is orders of magnitude more than tread rubber.
One of the things that stiffens the belt is inflation pressure - which is one of the reasons inflation pressure has such a pronounced effect on RR.
What inflation pressure does is press against the belt and resist inward (radial) deflection. Another way of looking at this is that inflation pressure pre-tensions the belt(circumferentially).
Conversely, rubber is highly extensible, so the same amount of stretch produces less tension force compared to steel. The net effect is that the tread rubber is a very minor player when it comes to overall package. This is not to say that the type of tread rubber doesn't have a measureable affect on RR, nor to say that small tread elements won't adversely affect RR, but these factors are overwhelmed by the effect the belt has.
So, you may ask, wouldn't it be better for RR if they used something stiffer in the belt? And wouldn't this also help tread wear and cornering?
Yup!!! Which is why steel is the material of choice. Early radials used fabric belts - 4 of them! Glass has been tried (shatters in cold weather!) Kevlar (Its strength is good for its weight, but you still have to use more of it to get the same stiffness.)
Why not more steel? There's an upper limit to the amount of rubber penetration into the steel cable and this affects the rubber to belt wire adhesion. This was one of the early lessons learned in the steel belted radial story. Anyone remember the Firestone 500, which was followed by the 721? The 721 cable was to address this penetration issue. Didn't work as well as was hoped, because rubber penetration was only part of the problem. Rubber chemistry was the other part.
Pretty much every tire manufacturer has settled in the same vicinity with regard to belt wire. So difference between RR of tires is more a function of tread compound than tire construction, simply because the construction differences are minor. Probably the only major construction difference is cap plies (nylon overlays - whatever you'd like to call them). These add considerable mass compared to the amount of stiffness.
But nylon has one remarkable property - it shrinks when heated. So a tire generating heat will have it's belt restricted the more heat is generated.
Nevertheless, adding nylon to a tire hurts RR.