@Black-Square I think it's something about the linearity of the rates.
What I mean is this: In the Turbine Duke, the prop goes from invisible, to moving at perhaps 50-80 RPM (visually) in a fraction of a second, even though the NP indicates it's still spinning faster than that. But, after becoming visible, it does indeed take a while to stop. The animation doesn't depict a smooth deceleration from disc to visible blades at all (to whatever extent that it should be). Also, the audio seems to end long before either the NG or the prop stop spinning. It feels like the audio should last a bit longer, and the prop should have a somewhat more linear rate of deceleration. (To whatever extent that is realistic.) This is true of the TBM as well.
For example, the spin downs of the Lancair Evolution and FlyJSim Dash 8 Q400 in Xplane 12 seem a bit more realistic, although even the Dash 8's props just suddenly stop at the end of their animation... So maybe just the Lancair? It's audio is also more consistent in its taper off. Of course, I'm open to hearing that the Lancair is the one that's wrong.
Finally, while the TBM 700 seems to take a while to feather its prop after shutdown, as per your linked video, the 850 and later models seem to feather much faster: https://www.youtube.com/watch?v=QfSDPwsIAlo