That looks legit. The only thing that doesn't seem like it would work is that 3.3V is not preferred (it expects 0-5V for a 0-100% control, so I would be losing the 3.3-5V power range).
Could I use a simple transistor to jump the PWM pulse to 5V? What type would I use for that?
Note that the suggested motor driver has Different PWM frequencies in and out. Input is essentially duty cycle.
I wouldn't go grabbing the motor driver though, because it does not look like the measured current is separately available.
If I'm not mistaken, you also need to scale the current of the new motor to the controller. i.e. A measured 20 A means 5A to the OEM controller. You might be able to get away using an Allegro hall effect sensor to measure bidirectional current, but then you would effectively need a bidirectional voltage to current controller. i=e.g. +-5 V gets turned into +-5A at the old voltage. Go to www.digikey.com and search for allegro hall effect sensors.
https://www.apexanalog.com/ makes a lot of high voltage/high power op amps, but they are extremely pricy. Dual power OP amps are available for use in bridge configurations. A single supply does bidirectional control.