ther are several problems involved

the formulas may appaer accurate as they deliver results in tenths of any unit.

Older rangtables or drawings may also show "pseudo" preciseness.

Many german rangetables deliver inputdata and results on millimeter-paper with line thicknesses equivalent to 5-10 metres(per second) drawed with a Standard(fixed sqrt) curveline lineal. So any of the input data is best case approximative and the same is valid for the output data.

much data in printed rangetables was interpolated from approximative surveys of trajectories.

If they used a numerical solution and the intervals of calculation were comparatively wide - the result may differ considerable from a calculation with the same mathematical model with same constants but smaller intervals.

I had the impression if you can reproduce a rangetable with a new calculated solution within 5-10m/s compared to the original rangetable you can use the calculated one. This is even possible without any "Mach based" dragfunction.

I have seen ballistic models (mostly for small arms) wich omitted the vertical speed component for calculation of deceleration of the projectile.

influence of stability and obedience (hope this is the correct translation for "Folgsamkeit") of rotatet projectiles change with density of air

(orientation of the longitudinal projectile axis to the Tangent of the actual trajectory point).

with the result
under(over)spun projectiles may have different trajectories compared to the "normal" one. And they may additionally differ depending on angle of departure.