My post has been edited to include the word 'hypotheticals'. There are more than one possible ways to interpret the crews 'decision' to allow autopilot to continue prior to auto disconnect. The release of control by auto was either 'anticipated' or a 'surprise', or something in between, and bears very heavily on what happened next, (or had happened 'already'). eg. The cause of disconnect by exceeded control limits of a/p? The cause of disconnect Unreliable a/s? If a/c was deviating for weather, was it because of turbulence at the moment, was it Radar data? Was it both escape and avoid? One strong possibility is that there was not turbulence that was a factor, and the crew was merely deviating having monitored LH, and/or Radar. That is supported by the continued use of a/p? This may also serve to focus more on an Unreliable a/s problem causing release of control by a/p?