And I remember a case (with an UM orig.) where the filament was completely stuck, it could even almost no more be pulled out by hand. All because the printing speed had been set far too high whereby the feeder had caused a lot of deformation and abrasion. Thus the friction was extremely high.
I stay with my theory that friction forces after the feeder are the central problem, and those before the feeder are neglectable, compared. Why: The brake effect of pressing material through the nozzle must be by far higher than the brake effect of spool unwinding, and these brake forces are the cause for the friction forces. Just normal engineering logic.
How much is the deformation by the knurled wheel, anyway? Could it happen that it increases the maximum filament diameter to more than the tube inner diameter? Then the friction would increase overproportionally. If the increased counterforce of the feeder wheel in turn causes deeper filament deformations (just a possibility, not an observation) then we have an escalating problem!