It’s very tight in there and must have been very difficult to install the mains CTs. Looking at the graphs, my suspicion lies with Main_B reading low as Apt_A never exceeds Main_A. while Apt_B always exceeds Main_B.
There are some other subtle clues. Most of the time the two legs appear to be completely independent, indicating mostly 120V variable loads. But there are a few interesting spikes at 01:00 (second one) and 09:00 where both of the mains spike at the same time. That appears to be a 240V load, but the two spikes are not equal in amplitude - Main_B appears to be about half the spike of Main_A in both instances.
It’s very hard to get a CT to read too high. Most of the time when they are off they read low. Moreover the Main_A is validated by Apt_A.
Another clue is that the power factor of Main_B in your original post is 92 where the Apt_B PF is 99. That looks like a predominantly resistive load and Main_B should have a higher PF.
All that suggests to me that the Main_B CT may not be properly seated, may have impurities between the cores, or may be cracked.
The picture doesn’t show the whole box and I can’t tell if the mains run down to the bottom of the box or out the side. I think they run down. If there is more room down there to install the CTs, I’d suggest at least moving the Main_B down there. If not, you have a pair of ECS25-200 clamp-on CTs. You could substitute one of those for Main_B to see if it resolves the difference.