Thanks Paul. I have seen that sort of passive resistor network, with various R values, and one that had a 1:1 isolation transformer after the resistors. My concern (without knowing if it's a valid concern) is would this sort of setup also sum the noise of the two channels? Depending on how the stereo tracks were mixed, would there be phase issues? Those concerns can be addressed (I think) by doing the summing in the digital domain, using some of the features of an app like Audacity.
The downside is that an Audacity solution would not be real time so not applicable to streaming from Tidal, etc; the resistor solution would be applicable there.