You're absolutely right that the issue is the coordination cost. It's not about the number of resources but just the fact that you need those resources to coordinate, slowing everything down.
Not a problem in waterfall, since you can set the targets beforehand and just have the teams work in parallel. In an "Agile" setting though (not the manifesto version, the current practice version)? It's a huge mess.
So then the question becomes, is this coordination cost worth it and how much are we willing to "spend". It seems the cost is worth it for Android and iOS since the native app experience on those platforms is so much better.
The macOS app is just a tweaked iOS app so it's also easy to justify.
But what about Windows? The Microsoft provided APIs are a disaster that keeps getting rewritten every half decade, and even Microsoft barely uses them. Windows users don't have a "feel" for a native app the way macOS and mobile users do, since there hasn't been a "native app experience" on Windows for a very long time.
Just look at the 2 context menus on Windows Explorer, they can't even get the sizes and colors between them right.
Meta would have been better off doing like Telegram and just using Qt.