I would argue that UE5 is even less suited to a game like this than Unity is. Unreal certainly has impressive rendering tech, and it has designs towards increasingly becoming a generalist engine, but it is clearly designed with certain genres in mind (i.e. 1st and 3rd person games like RPGs, Shooters, Action games, etc.). A city-builder in UE5 would present a whole host of other challenges, and many of the high-tech rendering features would likely be overkill. Not to mention, Unreal games have notorious performance issues of their own--though there is dedicated effort to resolving those.

Unity is designed more as a general engine, but it comes with a lot of baggage in terms of half-baked features and optimization difficulties. As the author mentions they really unlocked their potential with implementation of Unity's ECS framework, but they were still chained to Unity's rendering tech, which has been underdeveloped for several years now.

My observation tends to be that simulation games are the ideal case for custom engines. While there are some commonalities across games, compared to many other game genres, they don't get a lot of benefits from standardizations. Sim games often end up kneecapped by trying to conform to existing engine frameworks instead of spinning up something optimized to the way their systems work. It requires a lot more technical know-how than an action-adventure game or a platformer, and the up-front cost to developing your own tech is an order of magnitude compared to using out-of-the-box solutions. I think with the massive success of C:S, Colossal Order was in an excellent position to try something ambitious.

Maybe with open-source tools like Godot having more flexibility in their frameworks, where you can just get the parts you want (rendering approach, etc.), it'll be easier in future to develop more specialized custom tech for games.