To the best of my knowledge, SRA has never been thoroughly examined, and I don’t know of any paper in a peer-reviewed journal. In other words, no one knows what exactly happens when SRA is incorrect. In this Audio article the authors hypothesize, theorize, think that a mismatch causes IM distortion and scanning loss.
Azimuth: I know of only one paper that investigates the effects of azimuth:
Jakobs, “Analysis of crosstalk on stereo test records”, JAES 1971, p.280
One would have to examine the crosstalk figures in the light of perceptual thresholds
for crosstalk, such as from
Taylor, “The maximum permissible interchannel crosstalk for imperceptible restriction of stereophonic stage width”, BBC RD 1979/7
http://downloads.bbc.co.uk/rd/pubs/reports/1979-07.pdf
Klaus