Tool selection should be the function of not just the activity, but also the models

[david-istvan]

Example shown here: a1 and m1 implies t1 and f1, respectively, tau1 implements t1 and t2 <=> it's a good selection, but tau2 wouldn't be a good selection for a1, as it doesn't implement t1. (~Although the two tools support the same formalism, they support different types of transformations.) Tau2 is a good selection for (a2, m2) though.