I just tested some tweaked cyl trims. Took the base from the last blm stable ones at idle and adjust the off idle based on them. The engine started to run much better.
I think that if the trims are not optimal in the long run of engine running, some cylinders will get heated more some less and the total balance will be screwed. Also at off idle and take off conditions the engine will be heat soaked differently and it will also affect that and the engine will run like crap all the time even though only idle trims are off.

After good warming I checked the blm at CL on idle and they were almost even. Than decided to give some play with EOIT target.
Tried all kind of it from 40 to 70 in hex. |First I thought stock was $50. It turned out it was $60, Than start increment by 1 and at $63 what a difference. I looked at the afr, map and feel for reference, if AFR goes lower the fuel is evaporated much better and the engine have excess fuel leading to richer mixture.

I got lucky from the first try to get the sweet spot on the first try.

The fuel evaporation test with EOIT must be done at fully heat soaked engine and increment it by 1 and wait at least 30-60 seconds to feel the change.
I guess the optimal settings is cam, temp, head, valve dependent, and the the time needed for best fuel evaporation must be found by feel and experiment.

Than I give it a short drive.

Now what a difference from small eoit move $60 to $63 and part dialing the cyl trims. The engine got quieter by a lot[I could hear the stereo again]. It felt some much better[from crap running to almost stock like running].
The interesting part is that you can not see any significant change in the logs. It is all by feel. It started to shift much better also and the overall throttle response increased.

And that is with guess dialed off idle trims. I still haven`t figure out reliable method to get this dialed unless there is a driver with and a tuner in car. Getting some steady conditions and monitor the blm closely for splits.

The blms are great for
dialing bank to bank difference[split blms are almost definitely from bad ind trims.]

I also noticed that in closed loop only couple of cells are used most of time. I am sure this is not optimal and I am looking for some tweaks there.
Anyone got any experience with it.

Now the wideband at closed loop goes from 14.6-15.1 all the time at stoich. At idle that changes to 15.4-16.??. I guess this could be from low airflow or it could be from not optimal CL settings.