Binary ensemble score and prediction column relationship

So let's say I'm using WeightedAverage ensemble of two binary classification learners.
Each learner will produce Score and Probability column.

Since it's a weightedAvaerage we will learn weights for each learner let's say A and B.
Resulted Score value would be (A*ScoreA + B*ScoreB)/2.
For probability column we just take median so Resulted probability = (ProbA+ProbB)/2.
I'm not quite sure why we decide to calculate probability this way for all binary Ensembles (no matter is it average, min, max).

In all other binary learners Probability and score is related to each other via calibrator. So there is a function of score which produce probability, in case of ensemble it's not and I find it bit confusing.