I appreciate your conclusions, and frankly, you're probably right.

I'm still quite curious how folks evaluate between four completely separately but presumably functionally similar outputs and conclude that work tree 2 is the winner.