When you tandem two voice coils that are current limited the net torque is shared by both motors on the same axis. This works with one or the other feedback to prevent skipped steps. However, the strain or stretch due to accelerating load differential and friction must not exceed the fraction of step you choose to use for N= 1 to 16 or 32 otherwise you may have position errors across the axis. with null error at the home recal. end of the axis. You can observe the error independently to verify your tandem gear-tooth bands have optimal tightness and that motor-cable-bearing friction is low enough to prevent hysteresis at the unmonitored end.
Dual motors per moving axis is standard when driven from one end. Closed-loop stepper is not standard but was used once upon a time in old HDDs when the track density was approx 1000 TPI or so.
I have used an open-loop dual motor/axis at high speed (>1m/s) by using software that verifies the maximum acceleration, Ka and velocity, Kv that it can endure then modify the set limits for margin in the application. Then only some fault load condition or recal fault could cause a position error. There is a natural tradeoff between step N, torque and thus maximum acceleration so closed loop stepper servo does not offer much greater accuracy or speed with auto-correction with fractional N steps. Any full step slips mean the setup is suboptimal for max Ka, Kv with excessive disturbances. The videos I have seen, indicate a very slow response in seconds to large disturbances rather than expected step rates on the order of milliseconds and similar correction times of an analog BLDC motor.
The big advantage is the reduction of power dissipation in the motor when pushing acceleration to the limits of the servo mass assembly or also called Head/Arm assy. but this is only a factor when not accelerating.
If your software does not support these variables, you may want to ask, why not?