The idea has merit with buses because:

1.)a very large proportion of the fuel energy is lost to active braking. The same is not true with passenger cars.

2.) They tend to use torque convertors anyway. Most passenger cars no longer use fluid couplings.

A few years back I 'invented' a crankshaft eddy current brake (a poor man's KERS) and conducted a feasibility study to see how much I could raise the engine oil oil temperature through vehicle retardation. Using our beloved NEDC (it may be flawed but that's the measaure that car companies are interested in) I was very surprised to see how little the service brakes are actually used in the test. In the urban cycle, the vehicle is slowed simply by lifting off the throttle. The only significant braking event is at the end of the extra urban cycle. Mechanical KERS counts for nothing in this scenario. Whereas with a battery system, this saving is counted.

It also places unrealistic demands on a CVT. A lot of people assume that because CVT is continuously variable, tha it is also infinitely variable. The driveshaft spins at ~hundreds of RPM, A KERS flywheel spins at tens of thousands. This ratio is hopeless.

You also have a horrible task of running your KERS system in series with the service brakes whilst maintainingbrake pedal travel and feel and satisfying failsafe criteria.

Needless to say I abandoned the Crankshaft Eddy Current Brake idea very quickly!