Well, I have to say much has been improved, but some issues do remain.
(1) National focuses to declare war on nations need prerequisites. This is to avoid Germany DoWing everyone while losing one war already, for instance.
(2) The autojoin of faction needs to be removed, with the nation at war and faction leader both having a say. Current situation often leads to either global war or breaking M-R pact early as Soviets DoW someone who immediately joins a faction.
(3) While the AI is better, and it really does show, it needs to stop building divisions so much and keep about 25% of equipment & manpower in reserve. Right now, it will fight generally OK until it's equipment runs out, then the AI army basically shatters. In my playthrough above, at war with the Soviets, it looked like it would be a long, drawn out war. Then the AI ran out of equipment, and of course, the steamroll started.
(4) AI should also not attack so much where it is hopeless. It is a waste of equipment, especially as it's units get progressively drained. Perhaps in time, if things are not going well, switch it to "cautious" stance. It should also look at number of battalions to determine strength and not number of divisions.
(5) Division of Poland according to M-R pact should not happen if Axis and Soviets go to war in the meantime.
(6) The AI simply needs to reserve a percentage of it's force - let's say 5-10% - to defending his homeland regardless of threat. Landing by sea or air in the UK / Japan / Italy / etc is too easy because the AI doesn't keep any troops at home. While it will strat. redeploy & sail to defen, such troops arrive without organization and by the time they're ready you overrun the enemy. It's both silly and ahistorical, because there were reserve forces there ready to deal with any such eventuality.
(7) Battle planner needs to stop merging fronts. It is impossible to do a pincer movement using the battle planner, because it will merge the fronts and just advance along a broad front.