You don't understand that table.


Draw a sine-wave, and cosine-wave. Divide each quadrant by the number of micro-steps you set. Now step those waveforms by one for each step (in the number of micro-steps). That will describe the current driven into each of the X/Y coil (using the simple analogy of a 4-pole 90 degree stepper). Where as at 1/256 would present 0.006 x current on one coil, the other would be at 0.999 x current (just read off the sine-value and the cosine value along the graph, those represent the currents passed into each of the two coils). The extreme example of 256 micro-steps, applying force tangentially to the spindle, the 0.006 holding torque is in the one axis - the other axis is near full holding torque (0.999). the next step - the sine-wave value increases, the cosine-wave value decreases, and so on.This helps to describe the potential to lose accuracy with high microsteps as silly-small drive currents on individual coils cannot translate to linear motion under aggressive loads. If - "you think" the drive sends more current - to balance the forces across the coils to the rotor this would have to apply equally to both coils - and you'd quickly burn your stepper out.