Car shouldn't be listening to the MAF at all when it's idling, it should be in straight up open loop and working strictly off the IACV (which bypasses the MAF, doesn't it?) Duh, of course the IACV is post-MAF.
That said, my 97 had some freaky issues trying to idle warm that were corrected with a replacement used MAF (and started great if gassy cold), and I remember on the Audi 5000 that its warm start issues were partially because it was trying to use the damaged front O2 to attempt "more closed" loop operation instead of its cold start behaviour. My guess therefore is that the ECU just lets it do whatever on cold start (GAS GAS GAS) and then on warm start tries to be more subtle by involving a few of the sensors which eventually involves the MAF and the ECU makes some bad decisions leading to the car stalling.
I'm curious how a more modern Subaru would deal with the problem, considering cold start is a big part of Euro emissions standards now.