So, would these differences be due to the regulation circuit, the LA or both? Everything else is pretty constant. The testing methods, the environmental surroundings, the initial cell voltage, etc.
I think perhaps you need to increase the number of output tests you do (say 5 sets on each lamp in each A2) to see whether it's a variation in the batteries or due to the lamp or A2 itself. Did you clean all the contacts?
Those look pretty identical to me. Variations in circuit/batteries/lamp will probably give you a lot of variations in output and runtime. 5% variation is probably typical just in the circuit. Add to that the lamp and batteries and the variation stackup could probably go to 10%