It appears from the graphs at the end of the AAIB report that the HP must have sub-consciously moved the collective down every time he got a height check (and was always above the nominal glidepath) - the graph shows clear steps of reducing Tq and lever position during the approach.
This would account for the cyclic program reducing the speed to keep the desired RoD which eventually resulted in the high nose up, low speed condition (despite speed calls at 80 kts and 35 kts) that finally led to the VRS.
As I said before - it is so counter-intuitive to have a configuration where cyclic controls RoD and lever controls speed - how did anyone think that was a good idea?