Headphone Extension Cable

The only issue with headphone cables is that they probably share a ground connection. So you want to keep that resistance as low as possible to minimize cross-talk, which means paralleling ground wires. The ideal cable impedance will match your phones, and not higher. Higher impedance (low capacitance) cable becomes an inductive compromise, although neither is a big problem shorter than a kilometer+. Most cable impedance is 50 to 75 Ohms, a good match for your phones and a bit high (inductive) for 30 Ohm phones.
Lets look at this carefully. The capacitance is in parallel with the load, so if the amplifier has a low output impedance (high damping factor), it will have very little effect on the voltage seen by the headphones. Lets assume the amp plus long cables have 1 ohm of resistance, and model the capacitance as after this resistance, even 100nF will have only 120ohms of shut resistance at 20kHz, which forms a divider with the 1 ohm to drop the 20kHz response by < 0.1dB.

No remotely feasible cable will have 100nF even if 100's of metres long, this just is not an issue at all unless the amplifier has a very large output impedance or cannot handle capacitive loads.

If you really want low-capacitance cable one possible geometry is use a 34 way ribbon cable and use several end conductors for L and R and the middle 10-or-so coductors for ground, leaving a lot of unconnected wires between each signal and the ground, something like this:

LLL . . . . . . . . . GGGGGGGGGG . . . . . . . . . RRR

The multiple grounds reduce crosstalk due to resistive losses, the large gaps reduce capacitance. However inductance increases so you might have to check whether that's an issue.

For long cable runs there's a definite advantage to running 4 wire, not 3 wire, to eliminate the common ground crosstalk, but that requires the 'phones to be 4-wire themselves.

In practice I'd just run standard twisted pair for each side, commoning grounds if required by the connectors. Two twisted pairs(*) can be reverse-twisted together nicely - the result is flexible which is normally a requirement. The thicker the wire insulation compared to the wire itself the lower the capacitance of the twisted pair, but as I've shown this is not something to worry about.

(*) very easy to make yourself using a cordless drill and a vice to hold the other end of the wires.
Mitsubishi A1-N CPU can take signal over 1000 meters using ordinary signal cable.
AMD make 8086 @ 20 odd MHz inside.
It was in my molding machine, made in 1992.

So, how long is your cable going to be?
Unless you have many sources of interference, use ordinary cable passed through a braid, and connect braid to Earth at both ends.

And if it is going to very long, think of a wireless signal transmission method