henra
I could imagine it could potentially be related to the Reynolds Numbers. With some airfoils Cd decreases significantly with increasing Reynolds Numbers (and thus L/D increases) while others don't react much.
I see where you are coming from, but most of that improvement comes in the Reynolds Number range from wind tunnel to full scale. Once you get to full scale RN the variation is much slower, although there is always a reduction in skin friction over the whole aircraft as RN increases. So I suspect RN variations are not the answer.
abgd
Is it still true that L/D remains constant for an aircraft with a lot of parasitic drag? e.g. hang-glider, biplane with lots of wires...
I don't see any reason why it shouldn't be. That sort of aircraft has no Mach number variation to speak of; in fact they have not much Reynolds Number range either, so it would be difficult to see any L/D variation at different parts of their flight envelope.