They have to cater to a pretty wide set of users, and deal with a wide array of hardware and scenarios. UEFI, especially whatever random implementations of UEFI their users have, can't cover all of it. Addressing those needs currently requires something like GRUB (or a customised initramfs, which would be my preferred solution, but it requires more know-how), but GRUB effectively has to duplicate a large subset of the work that the kernel does, and inevitably (if only due to lack of resources), does so poorly, hence their argument that this is good for security: it's better than the status quo of GRUB. Indeed, if you have a UEFI firmware and it supports your use case, and it's well implemented, then this project is of no extra use (though it seems designed to just get out of the way in that situation and more or less just boot directly), but Red Hat's userbase does not entirely consist of people who are in that situation.