Thanks for posting the video! A couple of observations:
The first few manoeuvres you perform show the helicopter exhibiting static instability. After your roll input the helicopter doesn't show any tendency to want to roll back to the original attitude. Dynamic stability is irrelevant here since the aircraft is statically unstable.
The manoeuvre at ~1:40 shows the helicopter returning to the original attitude without pilot input, i.e. it's statically stable. It appears to return to "wings level" with only minor overshoot and is hence exhibiting dynamic stability also.
It's difficult to determine exactly what's happening in the remaining manoeuvres since these involve quite drastic pilot inputs.
It would be interesting to see:
1. A comparison between SAS on and SAS off
2. The aircraft response with shorter, sharper pilot inputs to mimic, say, a wind gust
3. Control inputs during coordinated turns (with SAS off, applying in-to-turn pedal seems to fix the issue for me)
I don't know enough about the Gazelle's stability system to determine if it's working as designed or not. From what you've posted, it seems to be a question of static - rather than dynamic - stability.