In the past using Siril, I have never made any distinction between images shot before and after a meridian flip - they all just get processed together and Siril rotates the later images to match the orientation of the earlier ones. I would then apply a background extraction to the final stacked image.
However, with the new ability since v0.99.4 to apply background extractions on the entire sequence, it seems to me that separating pre- and post-flipped images into two separate sequences (or at least treating them separately) would now make sense, as one would expect the background extraction to be different for these two sets of images.
Would this be correct?