Why is no one mentioning the software?
As soon as I heard about the feathering extension, my first thought was software failure. The pilot unlocked feathering, but didn't activate it. Maybe the activation was triggered by Mach 0.95 forces the extension mechanism couldn't withstand, but that's a mechanical design with decades (centuries?) of mechanical-engineering experience behind it, and far less likely to fail than software.
As a software engineer I'm all too cognizant of the non-linear failures it's capable of. On my first job, designing embedded software to drive a new piece of hardware, I was all too aware that I needed to look closely at my code before I suggested the HW engineer might step in. In my decades of experience since, that never changed.