They have been consistent since late 2017, when they began testing for Xanathar's Guide. The only untested Race/Class options since that time were for licensed properties where the IP holders (Penny Arcade & Critical Role) wanted the options to represent their brand: and those options have not been reprinted in other books.
This is simply not true.
None of the Theros races which were new got a UA, for example (AFAIK anyway, and looking back I can't find any evidence to support that they did). On top of that, it's an attempt to dodge the issue by limiting it to race/class - when they've been utterly inconsistent on spells/feats/etc. being playtested/UA'd, even when they have potentially major game impacts. I could probably find others if I kept looking, but that came to mind immediately.
In terms of an example of an Option that received mixed feedback and another go around, the Genie Warlock: surveys said that people liked the concept, but not certain elements. So they did another approach. That's different than an option which was broadly rejected, like the Truenamer Wizard.
There's no evidence to support claims of consistency here. They've been all over the place.
Most importantly, they can only do what's physically possible. If they playtest stuff well ahead of things, then it can be revised, but increasingly, they're putting out UAs so close to the release of the actual books in question that the only option would be to pull stuff and replace it, not to tweak it and try again.
That was certainly the case here. If the reponse was "good idea, bad implementation", and even what they've said doesn't really suggest it wasn't (nor that it was, to be fair), there was nothing they could do about it. They just didn't have time. That was a choice on their part, and not consistent with allowing good time like earlier approaches.
Other issues are things like the Dragonmarks system being approved by playtesting/UA, strongly, and then thrown out last minute in favour of untested trash races which don't support the lore (yet more untested races - so that's another lot for you, breaking your claim re: 2017 consistency) because of their half-arsed mechanics, which would have been roundly rejected by UA/playtesting (the uproar was significant, esp. about Eberron fans), or one of the designers indicating that the Sidekicks UA was extremely positively received, and approved of, then the actual Sidekick rules didn't reflect the UA approach, to the point where the designer seemed slightly upset/apologetic about it.
(To be completely fair, and I am trying to be here, I do not believe WotC would make the same mistake with Eberron if it came out today - I believe they'd use the Theros/Ravenloft model of "By default you get one of these abilities or a Feat" and do that with Dragonmarks. So I think they learned from it, but it's post-2017 shenanigans.)
They're not consistent. And maybe that's fine, but don't claim they're consistent. I've proven that they're not.