This is just a guess
If, as Atwater suggested, energy is transferred from the ground to the wrist by a series of acceleration and deceleration of joints moving from proximal to distal positions, my guess would be that the center of rotation is the thickest portion of the spine just above the pelvis. This portion is accelerated sideways towards the target at drive/stride. At foot strike (more or less) it is rotated towards the target and thoracic extension/contraction is employed to assist the internal rotation of the throwing shoulder. So at that point the shoulders are rotating around the spine but the energy path may be more up the spine and through the shoulder. The upper arm is moving at max velocity before braking at elbow extension (just before) to accelerate the lower arm.
Now I think many younger pitchers feel that the center of strength is located in the non-throwing shoulder and they initiate rotation by pulling or opening the non throwing shoulder. Your GIFs also give this impression the way the line is drawn. It is very cool. How did you do it?
However, it doesn't seem to me the thoracic extension can really be set by pulling the glove side. I think this is most apparent in the Ventura GIF. What pulling may lead to is a sliding of the torso and arm laterally and a subsequent dragging of the arm through the slot as opposed to a whipping of the arm through the slot.
I don't feel the trebuchet is a good machine to illustrate the throwing motion. A whip or a fishing rod might be a better tool. If the rod were jointed you could really model the throw better.
I do not really know though.
Is any of this related to your question?