Interference, or getting rid of it, is a black art, and it should always be remembered that in theory, there is no difference between theory and practice, but in practice, there is.
It could be that the stray voltages, if injected into the local ground line, could affect the signal voltages "seen" by the ESC. Shunting them into the "power" side of things, the voltage regulation can probably soak them up and lose them. Earthing the frame direct to the battery -ve with a heavy enough wire should, in theory, work (but see the top line).
An RF capacitor at the motor terminals is intended to soak up RF coming from the motor before it gets to the leads by providing a short circuit at RF. Fitting one at the ESC output just changes the shape of the ESC output waveform, rarely a good idea. ESCs often have either protection diodes built in that shunt out of voltage range spikes into the supply, or the output FETs, due to their construction, for such diodes.
Seeing as its an elderly motor, is the insulation still as good as it should be? This might need testing with a high voltage insulation tester - high voltage back-EMFs could be breaking odd spots of insulation down and escaping.
Another thought - presumably the field winding is being fed via diodes from the ESC output - it could be that these are isolating the winding from the supply, and an extra set of diodes might be needed to shunt the spikes back into the supply, but to both + and -. Or maybe a dustbin sized electrolytic across the field winding. The voltage will always be the same direction, and it isn't necessarily RF that is the problem.