From my reading of the TGM book, I gather the impression that HK believed that the club released in the downswing because of a change in direction of movement of the hands from a straight line path to a curved path, and that the endless belt model explains the release phenomenon. Conceptually, the endless belt model theorizes that the club will travel at the same speed as the hands until the hands enter the pulley section of the endless belt model. Then, as the hands turn around the pulley, the club will accelerate relative to the hands and this represents the release phenomenon. However, this conceptual model is dependent on there being a straight line path movement of the hands in the early/mid downswing followed by a curved hand path later in the downswing, and "reality" doesn't seem to confirm this fact. If one looks at the hand path of professional golfers, their hand path is always curved and there is no straight line section. Here are two examples.
Bobby Jones strobe photo.
Tiger Woods photo
I used the slow motion Nike commercial swing video of Tiger Woods to produce this image. The red lines represent the clubshaft. The yellow lines represent the left arm (and I only plotted the left arm's location in the later downswing so as not to have too many lines cluttering the image). The green line is a hand-drawn line that plots the sequential movement of the hands over time. Point X is the point when release of the club is definitely apparent, although there is seemingly a small degree of release earlier.
Note that the hand path in both these images is near-circular, and there is no straight line path. Therefore, "reality" doesn't support HK's endless belt concept as explaining the release phenomenon.
Surely, a much better explanation is this mathematical model, which demonstrates that the club develops angular acceleration at all time points in the downswing when the pull on the grip is at an angle to the COG of the clubshaft.