It's more numerically robust than acos() - we can't hit a range error
if we get strange rounding effects - and we've got the value of the
sin of the angle already (it's the length of the x, y, z part), so
might as well use it. As length is necessarily positive, atan2() will
give an upper-half-plane resolution, matching what acos() gave us.
This incidentally means that you no longer need to normalize() your
quaternion before you ask for axis and angle.
Task-number: QTBUG-114313
Change-Id: If3fa2b371c72991f1f8f151f78ef7f9180aa87cf
Reviewed-by: Matthias Rauter <matthias.rauter@qt.io>
Reviewed-by: Jøger Hansegård <joger.hansegard@qt.io>
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
820eff05f4