Agree that it's not so much stance width as where the CoM is relative to the GRF vector.
On the lateral plane if the CoM is below the GRF vector then we will laterally biased to the inside of the turn and will continue to move inside until something changes. Such as the inside leg not getting out of the way (unloading the outside ski) or the CoM moving above the GRF vector through angulation. When the CoM is above the GRF vector then we are laterally balanced to the outside of the turn and will move laterally out of the turn.
There's a little complicating factor in the pace at which the CoM/skis cross.
If done slowly with a tendency to re-center at transition before committing to the new turn then the biasing to the new turn has to be deliberate. Withdrawing the support of the new inside leg from a balanced position is certainly one way. (But if we are laterally balanced over the new outside ski, lifting the new inside leg will do little but take the ski off the snow.)
If done at a faster pace with no pause (some may call it toppling) then the CoM will already be moving inside the new turn. If we're laterally balanced to the inside then the CoM will accelerate inside. If we're laterally balanced to the outside then the movement inside will begin to slow down and eventually reverse. Which we are going to have to do to get out of that turn for the next. Rinse, repeat.
Biased to the inside.
View attachment 157966
Biased to the outside.
View attachment 157967