This is an issue which has been cooking in my thoughts for a while.
On older discrete power amplifiers, which have suffered failure, the main power transistors test OK on all usual devices, when out-of-circuit. But when in-circuit, the amplifier fails, and replacement of those transistors solves the problem.
So normal testing is NOT able to identify the failed transistor.
Why does this bother me, if all I need to do is throw in a new transistor?
Because the overall majority of these transistors are now NLA, and the widely advertised replacements are almost guaranteed to be counterfeits.
So if there is a way to positively identify one failed transistor, then I do not need to replace all 4 of them, with dubious units.