Plus it won't work (in the sense of saving blockchain space) for people who have few peers and tend to move money in one direction, e.g. a consumer connected to a "paypal"-like lightning service who makes purchases but doesn't sell online, this could be a significant portion of users.
It's a great extension to Bitcoin, but it shouldn't be confused with what Bitcoin does.
Unless I missed something, outsourcing watching the blockchain cannot be called trustless; they need to know what to watch. If your watcher collaborates with your LN-hub, you are doomed.
Here is my get-rich-quick-scheme:
* Setup a big LN hub.
* Setup a LN-watching service
Then for each peer connected to both, I filter those that:
* have a balance with <10% on my side
* have an earlier balance with >90% on my side
* haven't been seen for a week.
For those, I submit the tx of the earlier balance. Sometimes I'll lose a bit but mostly I will gain more.
Profit.
(e.g. the large parked capital requirements acting as a barrier to entry, fees - if bitcoin is to be a low-volume settlement layer - plus people wanting to keep spending money consolidated (i.e. opening as few channels as possible) while still getting the benefit of complete payment connectivity)