It's probably old wives tales mixed with something someone did once who thought he (or she) knew what the issue was. It seems to me that the "idea" behind this is to avoid creating "rogue" windings that might induce voltages in the search head receive coils that might desensitize the receiver circuits or cause false triggering.
The meat of my answer: -
There might be 2 to 200 transmit coil turns inside a metal detector that creates the penetrating magnetic field and, the current that flows in that transmit coil will be significant because of the use of parallel tuning capacitors local to it. The current that then flows down the cables to feed the search head will be significantly smaller in amplitude (due to parallel tuning capacitors keeping the bigger currents localized to the search head coil).
That significantly smaller current AND the fact that the cable is likely "balanced" means that extraneous magnetic fields from the cable will be very small.
Even if the cable forms several turns that are directly sat on top of the search head, the magnetic fields emanating will be synchronous with signals picked up in the receiver circuits and may only serve to move the "balance" point slightly. Most metal detectors produce audible tones and the effect of a slight change in balance can be heard and accommodated.
So, it's possible that some slight imbalance would occur that slightly favours discrimination towards gold and away from ferrous metal (or vice versa) but I seriously doubt that it makes anything other than really cheap and badly designed metal detectors have problems.