That's really good news. If the ferrite core really solves that problem, then there must be some really heavy pickup resulting in wrong step numbers.
However, the test with the concentric infill shows that it is a problem of the stepper motor and the e-coordinate respectively. Linear infill and concentric infill strongly differ in the acceleration situation (with linear, motion comes to a stop at each line end, well, almost - while with concentric infill motion is limited by the x-y-jerk). So far, everything looks consistent.