If threads were air tight it seems some would get notably hard to tighten in the process of tightening...I’m thinking this isn’t what’s happening.
There are engine assembly bolts (not Subaru) that require sealant when they pass through certain areas as well - because the threads aren’t sealed.
In theory threads aren’t engaged fully - one load bearing face of the bolt thread is seated against the facing load bearing edge of the receiving threads. It is not 3 dimensionally tight between the valleys and peaks. This presumably leaves a permanent spiraling air gap all the way up the shaft. Not sure what then happens at the bolt head surface but given most metal to metal areas get orings or gaskets I’m assuming it’s not air tight.
Thats “theory” - in practice, an infinite number of interacting variables, some variation of possibilities would present themselves.
So if you’re hearing this, no matter the cause, I guess it is right in the beginning when the head leaves the surface, the “potentially” sealed part and before the volume of the bolt hood shaft increases during removal.