I think we have a kind of consensus here.
Simple aerodynamic control may be possible, with careful design, but only at speed. Thrust control is possible but complex and heavy.
Aerodynamic control has the advantage of simplicity. Minimal changes to the design to make a symmetrical set of tail tins but it cannot be tested at low altitudes and the launch is liable to be effected by the cold.
Thrust control is complex and heavy but at least it can be tested at ground level.
I'm not convinced about a piston launcher as it still relies on a reaction force from the free swinging girder.
Would it help if the glider and rocket engine were separate sections so the rocket could drop away after use?