Agreed. Probably the best way to fix it is to simply skip printing the tolerances whenever "Theoretically exact" is selected. It seems to me, that there should not be tolerances with theoretical dimensions in any cases, classical or GD&T. Instead, GD&T rules and additional tolerancing should always be used for theoretically exact dimension tolerancing. And, since we now have a rudimentary support for GD&T boxes, this should be ok to be fixed.
https://www.globalspec.com/reference/43 ... dimensions