@CaiJonas the stuttering movements appear to be a feature of the moves generated. They are not arc moves, they are a sequence of line segments. Here's an example from line 2270 of your example file:
G1 X60.133 Y46.575 E.6074 ; perimeter
G1 E-.5 F3000 ; retract
G1 X60.509 Y46.911 F18000
G1 X75.371 Y60.871 ; move to first infill point
G1 X75.561 Y61.029 ; move to first infill point
G1 X75.728 Y61.145 ; move to first infill point
G1 X79.425 Y63.53 ; move to first infill point
G1 X79.396 Y63.585 ; move to first infill point
G1 X78.911 Y63.328 ; move to first infill point
G1 E.5 F3000 ; ; unretract
;TYPE:Internal solid infill
;WIDTH:0.420424
G1 F12000
M106 S81 ; enable fan
G1 X77.989 Y64.25 E.0389 ; infill
The commanded speed is F18000 but RRF can't maintain that speed at the boundaries between moves because the direction changes between the moves would cause the jerk limit to be exceeded. So RRF reduces the speed at the boundary to a value that doesn't cause the jerk limit to be exceeded.
It's possible that you are getting movement queue underruns, which would contribute to the jerky movement. The M122 report you published doesn't show any, but perhaps you didn't run M122 after a print.
If you are getting layer shifts then it's likely that you have the maximum speed or acceleration or jerk set higher than the motor can handle at the current you are using. You can try increasing XY motor currents, if the motor rated current allows. This won't help if the speed is high enough for the torque to start reducing. You can use the calculator at https://www.reprapfirmware.org/emf.html to estimate that speed.