I was really only thinking about enlarger lenses here rather than duplication/reproduction lenses optimised for a particular magnification. I’ve tried quite a wide range and I’m sure you have as well and they vary considerably in their suitability for film copying, but then that isn’t what they are designed for. Vlad’s Test Target is very good for assessing whether the compromises that you accept by stopping down are acceptable or not, and the main compromise would be to trade off central sharpness for depth of focus to allow for the curvature of film, and yes that will be a personal decision.
I suspect that lenses like the Sigma 70 & 105 ART have floating elements or lens groups to correct for different magnifications, much like the 55mm Micro-Nikkor from many years earlier otherwise I don’t see how they could be such good performers for film copying. Some of those $xxxx lenses have manually adjustable floating elements to do the same.
The 75mm Apo-Rodagon 2x does seem to perform well over a wider range than the symmetrical 1x version but the 1x is demonstrably better at 1:1 (those comparisons are on another forum). Still that lens is reportedly the same as the fixed aperture 75mm Magnagon in the Imacon ‘virtual drum’ scanners. They have a slightly different task in that these are line scanners and the tri-linear sensor is (from memory) about 43mm long but clearly that has to work over quiite a wide range of magnifications to accommodate 26mm wide 35mm scans at 6300 ppi to 105mm wide 5”x4” scans at 2040 ppi.

