I haven't tested this yet, but I think our assign_wells function is not able to allocate wells with zero filter length.
In this case iMOD5 directly allocates these wells to the layer, I have to verify if our function does that as well, but I couldn't find any specific logic for it in my quick glance.
Validation at least throws an error now, which it shouldn't.
Furthermore, ideally, I'd would like to align assign_wells with imod/prepare/topsystem/conductance.py. To support the 5 methods that iMOD5 to redistribute well rates.