I was able to reproduce at least a partial failure of the seat belt adjuster (i.e. got one of the two sides of the retaining wire spring to pop free of the divot on one side of the metal crossbar). To do so, I simulated the tail of the seat belt (the loose end) being trapped under the child's bum, and then applied moderate-firm force to the main part of the seat belt, pulling it towards myself (toward the back of the car) as if the child was leaning forward hard. The combination of this force with the seat belt tail being held trapped caused a torsional force to be exerted on the metal crossbar, which was sufficient to pop the spring free from one side. I heard a small metallic "ping", and on careful inspection it's visible (see red-circled area in photo below), but it would be easy to miss.
With one side of the spring loose, it may become possible for the metal crossbar to twist sufficiently to work loose, although I didn't observe that happening. It's possible that this is what happened in the original poster's situation, and when the sudden braking momentarily released tension, the seat belt adjuster came apart.
View attachment 70415
I was easily able to fix the seat belt adjuster by removing the cover and putting the wire spring back into the divot on the metal crossbar, but it's concerning that it's possible for the belt assembly to fail in this non-obvious way. I've definitely noticed my kids sitting on the belt tail on multiple occasions, so I don't think this is just a theoretical problem.
To the original poster -- if you haven't already, please contact the NHTSA, and let us know what they (and Tesla) say.