Actually the filter/mirror is constructed by vacuum deposition of many alternating layers of materials. The thickness and composition of the layers allow tuning the mirror to pass or reflect light based on frequency. (Dichroic Filter)
The incandescent filament produced about (5-7)% visible light, the remainder was pure radiant heat. The filter/mirror passes the visible, and recycles the heat back to the filament. The heat reflected to the filament replaces electrically produced heating, reducing the electricity necessary to make the filament glow.
I know, I browsed the paper.
Still, the effect of the composite device is that the radiation is no longer black body--as you say, it is concentrated in the visible range, because infra-red (most of the black-body curve at regular filament temps of about 2700 K), can't escape.
A brilliant (pun, sorry) idea. The practical questions, are cost of the layered envelope, and ultimate durability of the final composite structure. Will the filament hold up any better than it does with a traditional bulb?