playing with lovely mr Ink, I made my own small modification to the secondary rig scapula behavior, that seems to help. Just simply working towards an emulation of its sliding/rotation movement to help the overall behaviour. I'm sure you all have your own ways, I share mine (no ego involved).
A simple aim constraint (of the current rotational transform) to a transform sitting roughly on the imaginary tangent of the rotational/sliding movement. In spine4's space (parented/Constrained). I like simple solutions.
It's also quite modular-rigging friendly.
I can't seem to be able to post a movie file. I will at some point!