As the comment in your screenshot states, the queue is validated at withdrawal time, which prevents it from being used as a rug vector. In terms of DOS the queue will be capped after we push the changes from our latest audit. You’re correct the Vaults intentionally don’t handle fee on transfer tokens, but that has nothing to do with the standard or its viability as a reference implementation. Find the criticism about centralization odd because this is a pretty standard yield aggregator, we don’t have much more control over the system than any governed system like Yearn or Idle does, and again irrelevant to the discussion here. It’s provided as a reference impl to show how a complex system could implement this standard, not as a minimal base for other contracts, that’s the purpose of the Solmate impl. I’d be happy to side with its removal if you and others think it’s important.
I think there’s plenty of room for skepticism about the robustness of the Rari Vaults contract, which is why they’re not in production and one of the reasons why we made this standard (to reach community consensus on an optimal interface)! Would love to field more of your feedback but would request we do it outside of this forum if it’s about implementation details and not pertaining to the ERC4626 interface.