FTR...The actual computed acceleration value was 0.68 Gs. There is no point debating the precise accuracy of these numbers. The only important take-away is the fact that even if the 2 final BFO numbers are off some, say 20 or even 50 Hz due to unaccounted for error, the descent rates implied by the final 2 BFO values are very high, exactly as we observed in the simulator. In addition, via pvt correspondence with a high level official at ATSB, I learned that Boeing observed similar post fuel exhaustion descent profiles.
One more point: There is not a single piece of evidence or theory that has been suggested to explain why the final 2 BFO values would not be as good as all the others used in the analysis by Inmarsat, ATSB, the IG, and everyone else that has modelled the path. There were relatively small BFO errors at 16:00 immediately following that reboot (~17 Hz) and we suspect a similar level of error likely immediately following the 18:25 reboot, but compared to the ~250 Hz change at 00:19:37, a post boot up error of 17 Hz or so is insignificant in the bigger picture: It was descending very fast!