You're just concerned about unsprung weight? What about the now positive torque applied to those wheels? That's the whole basis of how ESP works: applying a counter-force to a wheel in order to balance the car thanks to the torque generated around the yaw axis, so how does each motor knows what force to apply so as not to actually throw the vehicle off balance?
Think RWD cars. Floor it on a curve and you can go into a tailspin. But RWD cars are designed as having power on the rear while FWD cars have not, so the effect could be much more dramatic with much less force.
It's not that complex to solve... as long as you have access to the various sensors (Hall, inertia... if any) and actual decisions of the ECU, and design data (i.e everything the manufacturer has), but piggybacking on it is quite an endeavor. And it has to be done for each car model since they all differ dynamically and in hardware.
I'm not even considering collaboration/conflict between ESP and the retrofit. A car whose chassis is designed to have force applied on the front wheels only (hence with rear wheels freewheeling) may not react properly at all to rear wheels having power. From suspension to weight distribution to force transmission through the chassis rigidity.
For example (oversimplification), FWD cars have their suspension balanced to be stiffer on the rear because it gives the front more grip (and the reverse is true for RWD cars). This stiffness would not usually result in a loss of grip under load (e.g in a corner), but adding a force to the wheel may overcome the remaining grip and throw the car into a tailspin.
Applying a rotating force on the wheels creates a pinching effect, momentarily changing the parallelism. This effect is sufficiently noticeable that some cars are designed with wheels not parallel at rest, so that when you drive on the highway (and thus apply some power to the wheels) the wheels are parallel.
There are countless scenarios where things can go wrong because you're simply doing something that the car was not designed to handle.
(For the curious, the Forza Motorsports series has a nice sandbox mode allowing to experiment with various settings by changing them on the fly, and the inline help explains succinctly the impact of each knob on your car's behavior)
But these electric engines could be programmed to only run when the car is moving in a straight line. If the wheel speed between the two rear wheels starts to differ then the motors turn off.
Think RWD cars. Floor it on a curve and you can go into a tailspin. But RWD cars are designed as having power on the rear while FWD cars have not, so the effect could be much more dramatic with much less force.
It's not that complex to solve... as long as you have access to the various sensors (Hall, inertia... if any) and actual decisions of the ECU, and design data (i.e everything the manufacturer has), but piggybacking on it is quite an endeavor. And it has to be done for each car model since they all differ dynamically and in hardware.