Which brings us back to my intuition-led hypothesis.
the thread acts as a physical restraint on the head, during the low-power warming-phase....we've already established the alloy head expands faster, but when the barrell is also up to the sama temp. the interference- fit is restored (within the plastic and elastic limits of the alloy, it being the softer, weaker material)
The wall- thickness of the cylinder, is effectively reduced to the thickness at the root of the thread...the thread is strongest at this point,being the base of an essentially triangular section....again, during differential expansion,the "overlap" of the threads, male and female, will diminish....in extremis, the alloy thread would ride up the flanks of the steel thread and either slip off, or the crests would shear each other off...essentially a broaching-action.
No doubt the size of the interference-fit collar has ben determined to give sufficient frictional grip to make the threads redundant under normal operating conditions.
I am an afficionado of Scott motorcycles....this unique 2-stroke twin has a central-flywheel with overhung cranks in 2 separate crankcase compartments.....the flywheel is furnished with a tapered bore each side...the crankshafts are pushed in, each having a matching taper...the assembly is pulled tight with a long draw-bolt.
These engines run sucessfully and dismantling often shows the tapers have cold-welded to each other...do not discount the frictional grip of metal against metal!
Aviation alert! The Scott engine was an option to power the Flying Flea (Pou de Ciel )