One of the destructive tests was executed. V_Drive was set to 48V, and a motor was attached with a nominal draw of ~10A while not under load. So when the bridge switches on, the inrush current should detonate one (or both) MOSFETs on the active path. Which is exactly what happened.

CMOS control logic survived the violence (and so, I'd figure the upstream control system would, also). But when the MOSFET blew, the motor was locked ON. Not only that, but it remained on when the control circuit was removed. Which I can't yet explain. I'll do a more complete post-mortem later. It may be that the explosion of Q6's pin1 caused a short elsewhere, as there was a large mess surrounding Q6.