One other way to solve this is to:

- use EMI sheilded conductors (at greater expense), or

- use twisted pairs where applicable.

I believe the CAN standard is based on twisted pairs, and they work very well for other ground/signal based configurations.

I had actually used EMI shielded wires - but, I think I must have had a faulty connection from shield to ground. Practically speaking, EMI shielding is hard to get right if you're not a pro. I only recently got an Oscilloscope and was able to measure what the EMI noise actually looks like on a wire. That helped a lot in getting rid of it, too. VFDs are pretty awful - the one in my machine nominally outputs 8amp,220v, 500hz power, containing that is a matter of extreme care. My previous CNC (now sold) really just didn't make any attempt and just counted on physical separation of wires and software filtering of signals.

CAN uses a differential pair which is the way to go. EMI will drive the pair in the same direction. Also used in USB, RS-485, pci-e, hdmi, etc.

You don't really have to twist them but it is best practice.

Had a similar problem on a stepper motor motion control system in a vacuum chamber. The single electrical feedthrough forced motor and signal cables to be in close proximity. The pwm noise from the motors would couple to the opto isolated limit switch signal and force the limit switches on. I solved it by putting 10k pull down resistors with a bypass capacitor to ground on all the inputs. Phantom stuck limits were no more.

I’ve found recently that ferrites really help with this. The electrical noise from large steppers is really erratic, but responds well to ferrite beads with 2-3 wraps. I put one at each end of the cable, since back-EMF is very real in large steppers.

Also heat. Wires bunched up tight can’t cool themselves down as well.

If heat is an concern, you'd probably want to spec a thicker gauge wire.

Regardless of wire gauge, bundles of wire noticably reduce your heat dissipation ability and therefore reduce your safe current carrying capacity accordingly. That's why electrical code significantly de-rates the current capacity of circuits which run more than just a few conductors in a single conduit or through the same hole.

Yes, but (especially when Cu and Al are pricy) being able to use a thinner gauge can be a big benefit.