I work as a designer of switch-mode power supplies. So fast rising current pulses and what-to-do about them are my bread and butter.
My boss used to say "Have Frequency, will travel" as a nod to a 50's cowboy TV series, but he's right. Steep edges have harmonic frequencies marching off into Poland.
Digital servos are characterised by having faster responses. This is achieved by using microprocessor controlled feedback along with very low resistance MOSFET drivers. The instantaneous current delivery can be an order of magnitude higher with a digital metal gear servo on 6V than with an analogue plastic gear servo on 4.8V. The significance of the metal gears is the increased moment of inertia and the much higher torque required to follow the "improved" position demand.
My suspicions would be that the pulse from the supply line to the servo is generating a signal in the signal wire, probably from magnetic cross-coupling. If that's the case, twisting may make things worse and ferrite will not help. In order to test my theory, you could separate the signal lead from the other 2 wires in the threesome. You may only need to move it a couple of mm. Of course if it works, you'll have spoiled the servo wire and if it doesn't work....well, you'll have spoiled the servo wire.