After countless flashes, I have arrived at this and can say that I am close to being happy.
The changes made are load scaling (with the kind help of @
Kodename47 ), OL/CL timers, OL fuel, and last but not least I changed the AVCS to start retarding the exhaust already from 3200 rpm. I also removed the OFH AVCS dip in the intake advance.
The AVCS changes have helped a lot on the AFR dip which used to be at it's worst at 3900 rpm (the yellow marker). It moved slightly lower in the RPM but is now nowhere near as bad.
The load scale change has removed any hesitation or choking up at low rpm. Note that the log is flooring it at 1500 rpm. Also, it was finally at reasonable IAT (starting at 23 deg C and ending at 18 during the pull). I still run the same timing and there is no knock correction worth mentioning. I suppose this should be relatively track-ready.
The AVCS changes have been made over 4 flashes where I gradually moved it lower in the rpm and compared with Virtual dyno to check for any ill effects. Besides from the car pulling cleaner in extremely low rpm I can't say there is much difference in the perceived power. Load, as an indication of torque, does seem to come on more linear and earlier (from 3500 rpm) up till the next dip at 4500 rpm.
If I wanted to go nuts, I could probably try to fine tune intake AVCS and see how that would affect AFR. But I think I am well past diminishing returns already. If anything I would probably be better off spending time rescaling MAF and rechecking the load limits. As long as my LTFT don't do any crazy things I won't be doing any more for now though.
Thank everyone for all your help and any further inputs are appreciated.