I guess that I need to start with that word: “ Stereo”.
Many people think it means two channel. Or sound coming from the left and right . It really means “solid”. The original Blumlein patents on stereo recording are concerned with using a configuration of microphones in such a way that, using only two channels, they capture amplitude and phase information so as to form an aural three dimensional image of a performance taking place before them. This necessarily included not only the performance itself but reflections of the sound from the boundaries of the room within which it was taking place thus creating a “solid” image of the event. A bit like an audio hologram.
It was literally about making a record ( in the sense of a historic report) of a real event in time. Of course Blumlein’s invention was before WWII when all recording was of a complete performance recorded direct to disc (or, rather, wax). Tape recorders and the ability to work outside of real time did not yet exist.
The reason that I introduce this historical aspect is that when commercial stereo records and reproducers were introduced post-war the same idea of the disc containing a record of a performance occurring in a real space continued for some years and related to all genres of music. High fidelity was originally developed to reproduce that performance /space as accurately as technically possible. A superb description of this concept may be found in the speaker positioning instructions in the user manual that came with the original Quad ESL in 1957. You are invited not to find the best place for bass but to envisage the plane between the speakers as equivalent to a theatre’s proscenium arch with the space behind as that within which the performers are situated.
Broadly speaking this concept still exists for classical music but other genres of music including much jazz have moved away from it. The main reason being the use of multi-track recording. The preference of producers for using this method is so that elements of the recording can be manipulated individually without simultaneously affecting other elements. Each element (vocal, instrument, group of instruments) is recorded to a separate track. For this to be successful it is essential that the sound of the other elements is not captured on that track. To do this the elements are screened from each other (called stopping or reducing “bleed”). The sound reflected from the recording room itself contains the sound of all the elements performing at the same time. This is therefore yet another form of “bleed” and is suppressed. So without the sound of the room no true “solid” image can be created. Hence a synthesised equivalent is created by the producer in the final “mix” to give an illusion of stereo.
This synthesised stereo illusion, effectively made up from elements recorded in mono, need not have any relationship to reality. Hence the listener may have no way of judging its likely correctness beyond generalisation. Therefore any spatial aberrations such introduced by upsampling will probably pass by the listener unnoticed unless they are gross.
In fact here in 2020 few rock/pop producers even aim for a synthesised stereo mix. They prefer to use a broadened central mono image with left and right margins for placing effects.
Live recording (concerts etc.) has its own technical problems. For example a microphone configuration such as a Decca Tree cannot be placed in front of the stage so as to obscure the audience’s view. The acoustics of the venue may also not be ideal. So, these days, a basically stripped down version of the studio multi-track system is used. Bleed is still a problem but some acts can go far to minimise this. I recall seeing Steely Dan on tour a dozen or so years ago and they actually used Perspex screens between elements of the band to reduce “bleed” (the Dan usually recorded their gigs). The end result is again not true stereo but a synthesised version. Of course although the public are unaware many “live” albums have been wholly or partly studio confections in any case.
Sorry this is long but it is why I hold that spatial oddities such as may occur with upsampling may be largely insignificant aside to those listening to classical music or to other recordings made prior to the late 1960s and 70s .