It doesn't seem like a necessary step really. The hub still needs a network connection to open channels. With that connection it can run some kind of SPV like c

It doesn't seem like a necessary step really. The hub still needs a network connection to open channels. With that connection it can run some kind of SPV like client. The bandwidth costs of that approach are larger, but still minimal. I suppose the one thing it does is make it easier for the hub to monitor for channel closure? But you could contract a third party service that messages you when that happens. Granted, that's a less verifiable trust assumption in a third party than they are broadcasting every block they get.