Yes, the 419mA vs 77mA is strange, maybe the regulator switched to a lower current because it was unable to maintain the higher current? Did you notice a significant difference in light output? See Dave's page about the
µCurrent, and the Silicon Chip article linked from that page, for more information about the effect of burden voltage.
Driven from a constant current source (eg. a lab supply in constant current mode), the lowest range including the measured value should be the most accurate, so for values < 400 mA, the 400 mA range should be more accurate than the 10 A range. The 10 A range will have a lower burden voltage, however, so for sources sensitive to voltage changes, especially a regulator that might drop out or shut down if the voltage gets to low, it may be superior. This is where higher resolution and accuracy sometimes comes in handy. This is essentially also what Dave's µCurrent does: use a lower value shunt resistor and amplify the signal to compensate.