If XFCE wanted to stabilize GTK, they could try to do that with a fork, but there was no bait and switch - it has always been the Gnome ToolKit. You should probably consider that before building desktops based off it that are not Gnome - its future developments are by Gnome, for Gnome.
If the fragmented ecosystems on the fringes of GTK - Elementary, XFCE, and Mate - were united, they could have easily forked GTK2 or 3 and maintained their own independent stable toolkit. Hell, the Gnome developers might have appreciated a wider swathe of developers working on it and switched to the fork with a transition to a plugin model for all the Gnome extras than constant churn in the toolkit.
Code churn for the sake of it can definitely be a hostile tactic, but GTK churns because Gnome is constantly changing at the whims of its developers in their pursuit of their own ideal desktop. Unless you are paying them for their time or contributing code yourself to their projects, why should you have higher priority over their own desires or those of their immediate users of their whole software suite?