One point on that. It was mentioned that the penalty for going over might be something like 5Sure. But I suspect that we'll want as many cities as we can "afford", whatever resources go into calculating this "afford". Because there are very few games, especially strategy game, where a player-controlled thing is not better than the automatic version of it without having some heavy penalty towards an excessive number of the player-controlled variety. I'd be very pleasantly surprised if deliberately *not* upgrading many large towns when you could do it will be a viable strategy. At best, my guess is that the best strategy is going to be to have a roughly fixed ratio of towns to cities (as of yet unknown) which might depend on civilization, leader, government, era, etc... but *not* on total size.
But I would like if the optimal number of cities was fairly constant between play throughs, somewhere around the 3-12 range. A "wide" empire would make up the rest of its size with towns, while a "tall" empire would have far fewer towns but about the same number of cities. This is because I want both tall and wide empires to be viable, AND to never have to micromanage dozens of dozens of cities. This is something that have a city cap could probably achieve this design goal, but a city+town cap seems incapable of.

So it could be (limit 3)
2city 1 town all good
3 city 1 town -5*3=-15 global happiness
2 city 2 town -5*2=-10 global happiness
So if you are wide (many settlements) you may want less cities.