You have neglected to multiply the results by the number of phases, but that results in calculated locked rotor torque much higher than the published value. The calculated value should be higher because the calculation neglects core losses and stray load losses, but it shouldn't be that much higher. Perhaps the published value is the minimum point of the torque vs. speed curve which doesn't necessarily occur at locked rotor. They may also be allowing for voltage drop in the power distribution system during starting.
Except for the missing multiplication by the number of phases, equation is as given by Fitzgerald, Kingsley, Umans, Electric Machinery 4th ed. It is derived from developed torque x speed = power dissipated in R2(1-s)/s. The power dissipated in R2(1-s)/s is calculated from the equivalent circuit shown in the question without the magnetizing branch (Xm & Rfe). Delivered torque is less than developed torque because friction and windage are subtracted.
The image below shows the torque vs. slip curves for both the rated and the locked-rotor values of the equivalent circuit elements. The values change as the rotor current frequency changes. The effect is a transition from the curve that is based on the locked rotor values of the circuit elements to the curve based on rated values. Sometime this transition results of a minimum value of resultant the torque vs. slip curve occurring between the locked rotor and break down points on the curve. See transition sketched in green.
