That’s strange indeed. It defintely looks like case 1. here, so a rebuild should have fixed it.
It also matches with the latest comment in the package AUR. https://aur.archlinux.org/packages/gzdoom#comment-904736