Thanks for verifying. Kernel workers are normal, just that udev-worker that never disappeared, as if a device was detached and re-attached in a loop, was not normal and caused quite some load. IIRC, on each loop, all udev rules were reloaded (parsed and in case applied again), which was the major reason for the load.
I remember. On a quick review of the maintainer scripts, I could not find an issue, but I wanted to add a debug flag to all scripts to find the command and condition which leads to the faulty dtb-* directory removal.