OK, so if I understand correctly, that “more complicated communication and syncronization code” is about how to organize that the server side works as a single threaded entity. If for example, you have multiple processes when horizontally scaling and so on. And another issue is how to handle connection breaking to the server.
But there is no extra need to do something fancier on the server. Like with OT, you have to transform things on the server. Here you still just have to decide if an operation should apply or not, and this is it.
That quote is just unclear a bit. Or am I still missing something? (I read both the guide and checked the colab code.)