Found this report, seems exactly the same problem that happened in this flight, although at a different flight phase:
autofeather with good engine
I can see autofeather kicking in way before any TQ going below 18% or any other engine problems.
We can clearly see in the diagram that ENG2 is unfeathering and starting to spool up just before impact (and so shows the video).
And Eng 1 turbine starting up (ITT spike) and other params increasing but Prop still very much feathered (or close to it). This would take longer because unlike the ENG2 that was never off, this was completely secured.
It looks like a few more feet, a few more seconds and they would have come out of it with two engines running normally.