Bsieker, a few counter-remarks.
1/ With the boost pumps working, the fuel manifold is under positive pressure. There is no suction in normal operation, except in the very short pipe from the suction point to the boost pump. This ensures that no underpressure and thus no cavitation occurs.
2/ The AAIB specifically mentioned that the boost pumps were working correctly.
Understood about the positive pressure from the boost pumps. However I'm not sure what parameters are recorded in the FDR for the boost pump. An underperforming boost pump (underperforming due to cold fuel viscosity perhaps) could create a vacuum at the outlet, if the HP pump was trying to move more volume. However if the boost pump RPM is monitored and recorded and shown to be within normal parameters, then I agree with you.
Originally Posted by AAIB Special Bulletin S3/2008
Detailed examination of the fuel system and pipe work has found no unusual deterioration or physical blockages.
Just fine parsing of the text, a collapse/rebound event might not necessarily show up in a deterioration/physical blockage detailed examination. I really do think you'd have to specifically go looking for the kind of evidence this type of event would create. But to be honest, I don't even know if the materials necessary for a collapse/rebound event even exist in the pipe work of this fuel system.
3/ Even if we did assume suction-feeding, the highest suction under-pressure possible happens when the engines run at high thrust with no boost-pump operating. This is within design specifications, expected, and will not cause a significant deformation of the pipes, sufficient to restrict flow to such a degree as happened in this case.
You assume that the design of this fuel system is sufficient for all "normal" operation conditions. However it's possible this is not the case for the specific conditions of this flight, and I conclude this mainly on the evidence that the same event happened on both sides of the airplane to 2 independent fuel systems.
Remember also that both engines initially responded with an increase in power, then both rolled back at slightly different times. Could the flow rates between the boost and HP pumps have been different enough that a vacuum developed over a fairly short period of time between the pumps? Might be plausible in theory anyway but could also be a real stretch.