Leaving aside the details of any mishandling of the fuel and fuel system, the reasons for which I doubt we can ever know without full data and voice recording (don't hold your breath) I think the final trajectory of the flight begins to make more sense in the light of the recent report.
The first thing that struck me was the Nr warning described as "intermittent". That surely is significant. An intermittent Nr warning suggests Nr alternating just above and just below the min limit (whatever that is). It does not seem to indicate a single failure to reduce collective, ie a failure to establish an autorotation. It is far more characteristic of someone milking the Nr for range to the limit.
Further, if one assumes that control was lost (not, I think, an unreasonable assumption) the only significant parameters affecting the subsequent trajectory are drag vs. gravity in the vertical and the previous heading in the lateral.
From this I suggest that the final track of the helo is that which the pilot set up/was left with when control as lost.
From the data in the AAIB report it would seem that they'd just crossed a park/were just crossing a park on a roughly westerly track and closing on the river which to my mind would offer the only realistic landing space available. If at 1000' as radar suggests over the park or just before it a jink left then tight to make the river might well be too ambitious as well as ending up in fairly a fairly congested part of the Clyde, bridge wise. However stretching the glide straight ahead to aim between the bridge by the Clutha bar and the pedestrian footbridge some way beyond would look much more feasible. The trajectory the aircraft took would offer a good shortest track onto the river, but if the range was just too much and the NR couldn't last their resting place is a logical place to end up.
Odd noises as reported. heaven knows what a helo sounds like becoming chronically overpitched in flight. I don't doubt it is odd, even to the public. As the overpitching would be likely to occur in this (hypothetical) scenario later in the event rather than earlier as desperation set in it would account for the relative integrity of the airframe as opposed to a freefall from 1000ft which would surely result in total destruction.
I thus suggest an auto entered acceptably, a decision that max range technique should make the longer stretch of river beyond the Clutha bar and a decision not to 180 into the park or jink into the river to the left with close set bridges, a later realisation that they were falling short, milking Nr to stretch and eventually in desperation slipping below the 85% range Nr to the non recoverable figure of - what is it - 83%? and losing it.
Even so, I doubt we'll ever know. Of much more concern are the anomalies with fuel quantity, warnings and the configuration of the fuel pumps which we've repeatedly been told simply couldn't have happened, yet clearly did. That would seem to suggest Human Factors as the focus, as ever.
If this accident doesn't bring in mandatory data recording I'll be very surprised.