I haven't worked on any compositors yet, but I do know there are several factors that control the modifiers (since they are just an optimization for copying the buffer efficiently), and many of those cannot be determined at the swapchain creation time. (such as the position of the window, the window having focus, subsurfaces on top of the surface, etc) It is perfectly reasonable, IMO, to be suboptimal to begin with due to the many unknowns.
I'm not saying it's not reasonable, I'm saying that if applications don't care (especially closed source ones which don't care about being good citizen) and they recreate their surfaces when they don't need to, it's not really up to Wine to mitigate that. IMO the WSI should be aware of such issues and take them into consideration in their design, regardless of what is reasonable or possible by spec. Fwiw you could (and possibly do) very well have Linux native games doing the same thing, and Wine wouldn't be able to help in any way here. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/11342#note_145150