Fix reporting of before/after to remove positional bias

[russell-taylor]

Consider that the reported score difference will include a bias factor for every Mover's coarse position, so will not reflect the true energy difference between the conformations. We'd like to add them when we're optimizing but not include them in the initial or final reported score calculations.