Because you are only looking at the data the way you want to see it.
They were very vague on certain information and completely left some out.
I have posted additional data on this thread from the raw ADS-B data.
When the AP disengaged the plane did not "flick roll" as you put it.
That is what the ADS-B data shows, the initial upset started with a 8 degree heading change in 0.156 seconds, (which is a rate of 52 degrees per second). The change in headings and altitudes and rate of change can and were calculated by me between ADS-B data points.
They say the "aircraft rolled to 45 degrees" That would easily have been caused by having 1 throttle at climb power and the other at idle.
That is partially the cause, the main cause is the autopilot would have been making significant roll input to counter the yaw and to turn onto the requested heading of 075, and that roll input is neutralised when the autopilot disconnected. So the sudden removal of the aileron input is equivalent to a sudden roll input at 270 kts., the ADB-B data shows corresponding high turn rates.
They also do not say when the crew eventually brought the right throttle to idle. Why not? They have that data, but they didn't release it.
They have that data and it was in the draft of the preliminary report, and subsequently removed before publication.
The last aircraft coordinate recorded was 5°57'56.21" S 106°34'24.86"
That wasn’t the last point transmitted, the last ASD-B point transmitted is just 78 meters from the main underwater debris field.
What happened in those last 20 seconds?
I have posted the ADS-B data for the last 40 seconds on this thread, you can see for yourself, look at the headings and the elapsed time between points.
“Don't be a show-off. Never be too proud to turn back. There are old pilots and bold pilots, but no old, bold pilots.” E. Hamilton Lee, 1949