Bearfoil,
The fundamental issue is that the PF applied mostly nose-up inputs - hence the THS progressively trimmed up (by integrating his input).
If he had applied the correct stick-forward action in response to a stall, we wouldn't be talking about this at all. The design of the auto-trim is not the cause - the THS acted in response to the PF's input. His inputs were not appropriate. Cause and effect.
In fact, if he had flown pitch and power, it wouldn't have stalled at all.