What would it cost to design in a supply-rail fuse for the ELT battery? Or failing that, a length of fusible PCB track.
Initially - practically nothing, retrospectively - quite a lot! (re-design, re-qualification, recall.....)
However, as I said in earlier posts Honeywell have been at this game a very long time, I really can't believe they didn't think of that and the need to protect the cells from shorts 'upstream' of the PCB(s), so the question we are left with is:
What failure mode did the box and it's installation manage to come up with that meant it didn't work?