I agree, it is a bit confusing - and as you’ve seen, isn’t quite optimal yet.
My concern with having an ability to migrate entire communities, is that the hosting instance takes the brunt of the load of the community. I’m envisioning a scenario where, let’s say Lemmy.ml is hosting a number of large community, and one community explodes - say, “funny.” Say it gets absolutely massive. Lemmy.ml is running out of capacity and either has to grow vertically, which is a cost to the owner, or it can migrate a community (or cluster of communities) to another instance that has the capacity to host it - this distributes the load.
Ideally, this will not be an issue for many years. Maybe the backend can be written as a sort of “raid5” across cooperating instances, both for load distribution/balancing and for high-availability. Who knows, I’m not a dev, and I have no doubt it would be a massive undertaking. But it’s not difficult to foresee problems down the road as the fediverse grows - especially if/when there’s a massive Reddit migration. 🤷♂️
Not that I can see, yet.