Well, it depends on what the editing principles were. What I mean is that it is quite possible that they have been both well edited and that such inconsistencies have been left unchanged. In fact, such inconsistencies are precisely one of the criteria for deciding whether a text is a composite. MN 111 is an obvious example: the text duplicates some of the jhāna factors. This must have been fairly obvious, yet they decided not to edit it out.
But to be a bit more specific, we find an inconsistency similar to the one at DN 21 at MN 47. Here the Buddha is first referred to as “the Tathāgata”, then as “this Venerable.” Apparently the version in Chinese has “this Venerable” throughout. It is hard to imagine that this obvious inconsistency is simply an oversight. I would suggest editorial policy is the most likely explanation.
But don’t you think they would have needed an excuse to include this in the sutta? Drawing on “literary antecedents” is what you would expect in the Jātakas, but not in the EBTs. It seems to me that the EBTs are generally limited to real events, or what was perceived as real events. Perhaps this is an exception, but how can we be sure?
Well, the sutta itself says nothing about anāgāmīs (non-returners), but rebirth in the brahmāloka (the brahmā world) seems implied by the word “mind-made”: “having been reborn among a certain group of mind-made [deities] that transcend the company of devas that subsist on edible food.” Moreover, the description “accomplished in … wisdom” normally refers to stream-enterers. (In fact I am not aware that it can refer to anything else.) And any ariyan who gets reborn in the brahmā world is at the very least a jhānānāgāmī. (These are non-returners because of jhāna attainment, not the usual kind of non-returners who have eliminated the five lower fetters.) So it looks like case 3 is possible.
It seems to be commentarial. Yet if the suttas limit case 3 to the jhānānāgāmīs, then this very close to the commentarial position.