I.e. server side decorations can be used as a primary option and if it's not available, Wine can fall back to libdecor.
That is how it's intended to be used, yes, but it is a workaround for applications or libraries that can't provide suitable window decorations themselves.
The idea that it's a workaround for GNOME isn't wrong, but it isn't right either. There are plenty of reasons why window decorations might not be supported, be it because the compositor needs to be as fast as possible, or it's just not technically possible.
I'm purely proposing libdecor here so that the experience on GNOME, Weston, and whatever desktops result from this migration and don't have xdg-decoration are able to integrate more "nicely", being subjective.
Again, I'm fine with it not having libdecor. In fact, using a library like that might even pose some technical challenges with things like web browsers running in Wine, which like to embed window decorations *in* their window, like libadwaita likes to do.
The performance hit with libdecor is more theoretical than anything, and would need to be confirmed before using it as a fact. It might even be fixable with some consultation with GTK folks on how to work with subsurfaces in a GTK context.