He was evidently not satisfied with, since he suggested the Te Deum as a substitute finale in perfoermance. [...] The only thing we know about Bruckner's final draft is that he was not satisfied with it.
That is not quite right. It is not that he was not satisfied with it. The problem is that he needed a couple more months to finish it. There are a number of passages that he simply hadn't composed yet or hadn't orchestrated yet. The reason he suggested the Te Deum as a finale (an inadequate substitute at best) was simply that what he had of the finale at the time he made the comment was very much incomplete.
When I hear one of the completions I hear two things: first, each attempt at a completion arrives at a different solution, therefore no-one can claim to present the composer's thoughts with any certainty, let alone any legitimacy.
At this point there is a lot of overlap between the few most recent completions because more original material has been unearthed. The large chunks that Bruckner had already fully orchestrated are the same.
Second, by the time he was working on the movement, Bruckner's language seemed to have evolved to the point where it couldn't follow the others without jarring results. Harmonically it doesn't sound tortured and ambiguous, orchestrally it is much sparer than the first two, and rythmically it isn't like any other Bruckner finale. Until the coda arrives, it sounds more like the first movement of an unwritten 10th than the last movement of the 9th.
I'm not sure I agree with that either. If you had said that the eintre 9th symphony is quite different than Bruckner's other symphonies, I would somewhat agree. But with the caveat that the versions of the other symphonies that we are familiar with are massively revised versions. If you listen to some of the original versions, you'll see that the 9th is a part of that same langiuage, albeit quite more advanced. The main themes in the finale exhibit typical Brucknerian rhthms and the development of ideas is very much in line with what he usually does. What is very unusual is the beginning of the finale. But then again, how else would you do it? Given where the Adagio leaves off, these gasping, groping, tentative sighs that open the finale are just about the only logical next step. Where we really have little clue as to what Bruckner wanted was the coda actually which was supposed to be this mega fugue of all the prior themes of the symphony plus some key themes of the 5th, 7th and 8th.
Again, a large portion of the various completions is original material, fully orchestrated. What Harnoncourt has the VPO play in his lecture are all the complete surviving fragments that were known at the time of that recording. That is quite a substantial bit. As he explains in the lecture, there are a few developments where we don't know how he wanted to modulate from the key of one segment over to the key of the next and then the coda is a big unknown. The main segments are otherwise quite done. The problems of the earlier completions was on the one hand a lack of some of these fragments that were subsequently unearthed and secondly overzealous editing by the person doing the completing, who often "cleaned up" harmonic dissonances and beefing up the orchestration etc., which is why the version Talmi recorded sounds like some Hollywood kitsch and not Brcukner.