The image below is way simplistic, but I think it will help some.
Positions A, B, and C are the same person in different parts of the transition. Position A is just before the release. Position B is when the skis are flat to the surface of the snow and Position C is when the committment to the new turn is infull swing. Of course during this process the skis wil also start to redirect down the hill but ignore that for the sake of simplicity. To start, ignore the red arrow. In position A, let's say that the shouldlers are distance X off the surface of the snow (as measured perpedicular to the slope). If the body is kept the same length, then the distance X will increase at position B. There wil be a nominal up movement. If there is an actual increase in the length of the body this up movement will even be greater.
Given a consistent turn radius/energy, my suggestion was that the shoulders follow the red line. The only way this can hapen is if the legs flex to shorten the length of the body. From position A to B this occurs by flexing the old stance leg. It is this movenment in the old stance leg that brings the skis flat to the surface of the snow. The same leg will continue to flex more to draw the CM into the turn as it develops and the new stance leg will legnthen as the turn progresses.
In this way, the movement is the same in the top of the turnas it is in the bottom. The leg towards the intended direction of the turn relaxes/flexes to start and increase the turn.
Because people, in general, tend to orient themselves in relation to true horizontal, when skiers are in position A, they feel like they will have to suck up (collapse) the old stance knee into their chests to get the CM to follow the red line. I feel like I wil have to collapse and fall down the slope as I try to collapse my CM closer to the snow than where I start. That is what it feel like. Of course the feeling lags behind what actually happens, that's why you try to go beyond what you think is comfortable and even then, it usually isn't enough.
Now my pooint about lengthening the stance leg through the turn and standing on it is for a lot of reasons, but here, it is so at the end of the turn, there is some room to flex into the next turn. Lots of people flex both legs the most in the bottom of the turn and then there is very little left to flex in order to release. As a result, the most common action is to extend the new stance leg right onto it's BTE. I always like to say that the flexion of the release in the old stance leg comes in diminishing returns. The biggest benefits come in the first 25% of flexion. If you have already used that up you have to do even more flexion to get the same release effect.
Another point here. Like many people, I also tend to flex my stance leg when the free leg flexes and lose some power in the turn that I could have -- this is also more tiring. So, I have to consciously think of allowing the stance leg to legnthen. It is more of a feeling of "standing on the stance leg" to pressure it than pushing on it.
There is a big difference between the stance leg lengthening because the free foot/leg action moves the CM inside the turn and pushing the CM inside the turn by extending the stance leg. These are two opposite causal chains and they have very different overall effects. The former keeps the stance leg passive, the later tends to introduce all sorts of bad stuff into the system that is unnecessary and inneffective.
I hope this is clearer.