If used while charging a lead-acid battery, then one of the LED would get quite high currents depending on color and type so great care would be needed. Some stupid car models may actually charge with 16V (owners of such cars can recognize this by them chewing through lamps madly...) and every V would increase the current with nearly 10mA.

If only used after charging (and a bit of rest), the variance for LEDref wuld be about 0.85 to 1.05mA (give or take a bit depending on LED color - 12.5V to 10.5V and 2V over LED). But it will take quite some time after charging (and no load) until the battery have stabilized at 12.5V. Not allowing any rest time, the variation would be 13.5V (1.15mA) to 10.5V (0.85mA).

But the variation in LEDref would still not matter much since the current variation in LEDvar is 100 times larger. No problem for the eyes to see the difference between two LEDs with so great difference in current between them.

So in the end - the variation in LEDref doesn't matter. The only thing that does matter for this design is that the LEDvar is living an unsafe life if used while charging the battery.