You probably are having difficulty creating an isolated measurement sweep sent to the mains and another one just to the (non-LFE sub) with a proper acoustic timing information anchored to one of the mains -- this method also requires a channel remapping capacity which is easy enough to do in software, but I wouldn't know how it's done directly in an AVR. A workaround would be to disconnect/power off the sub while measuring either one of the LCR channels; then, performing another sweep sent to the mains with them disconnected/powered off while the sub is active or turned on -- and just manually apply guestimate time offsets if examining the overlayed IR/phases.
Anyhows -- skipping all that -- and judging merely from what you've got right now (magnitude only curves) in the above graphs, it doesn't look like there's any obvious abnormal phase induced cancellation between the BM sub + mains channels from a non-optimal BM processor setup -- this was speculative on my part since I have no clue on how modern AVRs fair at coping with phase anomalies that could occur during BM, but it's maybe fair to assume some are also smart enough to have some tricks up their sleave to avoid/fix such issues. The dips at 60 and 90 Hz look like they're coming inherently from the room itself as they're common with the sub-only and LFE-only curves. Other than some very minor peaks and dips (which are likely perceptually insignificant, say, if you would apply psychoacoustic smoothing) that could maybe manually "corrected" a tad bit if desired with additonal PEQs, well, I think you're good!
