Considering it is a minimum 20ms for our brains to interpret a delay
Perceptible delay is not really the relevant measure here. In M/S processing (or any kind of processing which involves summing) a substantial delay will create comb filtering.
The measure of "substantial" is how many degrees at a particular frequency. A fixed delay will be a higher percentage of a full cycle at higher frequencies, so generally for worst case you would consider the effect at 20kHz (even though for most adults behavior at 20kHz is not particularly relevant it gives a nice consistent worst case target to use).
The propagation velocity at low frequencies is not quite the same as at high frequencies, but for a quick analysis you can just assume it is the same, and then dig into more precise details if it is within an order of magnitude of something you care about.
Propagation velocity in twisted pair is around 70% of free air, so about 200Mm/s, or equivalently about 5ns/m.
Period of 20kHz is 1/20k or 50us.
So the difference in propagation delay between 6m and 0.3m is the propagation time through 5.7m, or just under 30ns.
30ns/50us is 0.00057, or 0.057% of a wavelength. One degree is 1/360=0.28% of a wavelength, so the difference in propagation delay through an extra 5.7m of cable is around 0.2 degrees. And of course relatively less at lower frequencies.
I don't have time to whip up a Matlab script that shows the effect of summing two identical 20kHz sine waves with a 0.2 degree offset, but needless to say it will be very small.
In a real system that will be down in the noise, and for a M/S setup you should not typically have the M and S signal at almost the same amplitude in both channels, so any combing will be even less if the signals are not identical amplitude.
A little bit long winded, but an example of how you can think through questions like this and get a little better understanding of when you should be concerned rather than just looking to someone to give you a yes/no answer.
)
