I've ran tests and combination has no impact on SA/NP.
My feeling is that it also has no impact on the number of regular attributes but Al will be testing that from what I understand.
But am I reading correctly that you went from non-optimal 64-65 greens to optimal training 55 greens? so there would be a difference.
You should definitely see the difference between a non-optimal training and an optimal one, unless even in the non-optimal one the majority condition loss is from the world class drills, then a bit of non-optimal would be hard to spot.
What I'm saying is that SA/NP run like clockwork, they increase every x% condition used, and any change in the training level should be visible.