@Ian W
Well, the securaplane lab building burned down once while testing (without BMU connector plugged in)
Seriously, they surely did not test a cell to failure inside an airframe, nor with _all_ systems operating on a test bench.
System level testing is whats lacking, both during design and in production, for example, parasitic connections where found after TAK incident.
They apparently drove a nail into a cell. But they did not deteriorate a cell to short itself "naturally" by over/undercharge.
A nail through all layers of all sub cells is 10s of times better short than a single spot of a sub cell shorting!