The positional accuracy of the motors aiming the lasers (relative to the satellite orientation) probably comes from optical encoders. I'd expect servos in an active loop as you mentioned, rather than steppers - which tend to settle at 1.8* increments.

Either way I have no idea how they work out the firing solution for the comm lasers with sufficient accuracy. Hypothesizing: maybe they shoot multiple lasers, the main comm laser and a alignment pattern with auxiliary lasers which could be used for targeting feedback?