I'll look at the article, thanks.
I have not run the numbers myself for audio, but have seen thermal distortion from resistors in simple op-amp circuits when you look far enough down (in my case of a few years ago, 16 bits). If you have say a gain of 4 (12 dB) in an opamp, the output resistor sees 4x the voltage swing that the input resistor sees. On average, this just affects the gain, but for slowly varying signals you can see the nonlinearity caused by the gain changing with resistor (and opamp) thermals since the output R changes more than the input R, inducing thermal-related signal nonlinearity. This happens in low-level signals -- it's a problem of a nonlinear resistor ratio due to thermal tails.
For most of my work, the resistor thermal time constants are in the mud, so I choose low-Tc resistors primarily to reduce gain and offset errors. However, in some situations where the DAC switches and dwells, resistor (and opamp, and even wiring) thermal tails are a real problem. I verified the Rs were part of the problem with a simple divider to take out the opamp.
Without knowing the actual signal levels and resistor's (and opamp's) thermal characteristics (tempcos and time constants) I cannot say whether it matters or not for audio. Not saying you are wrong, just that I don't know, and when I have some spare time may try to dig deeper. My past experience may have misled me.
It is interesting to note that years ago those little metal resistors everyone wanted to throw into circuits for their better matching and lower tempco were actually noisier and had higher self-capacitance and hysteresis than standard carbon film. One of those trades one has to really look closely at to ensure the right film is used...
Thanks - Don
p.s. In present day, semiconductor thermals and self-heating are a HUGE problem! Smaller devices means greater thermal density, faster thermal time constants, and much greater nonlinearity due to thermal effects.