Adapting the guide: rerun with adjusted flags doesnt work

The readme.md and the .pdf still gives the advice that you could run the multi_Class_Sampling with some other flags again after it has crashed. This doesnt work any more. One the one hand when you rerun multi_Class_Sampling it asks for an empty "Results"-folder, but on the other hand it asks for the last threshold_step_0x when you try to start where it crashed with the flag

HaSa::multi_Class_Sampling(
    ...,
    step = x,
    ...)

in the exact same folder. So that excludes itself.