I think I can explain what is happening in the case of splitting an audio file into smaller files by CD track markers.
The problem happens if the user selects the option to name each track by its marker name and one of the markers has no name.
If one of the markers does not have a name, then Wavelab gives it one: the name of the original file itself! It then tries to save the newly-split new file with that name, and this creates a naming collision. Depending on how the other auto-split settings are arranged, there could also be a new folder that is created with the same name.
I think that an easy test would be to load a .wav or .aiff file into Wavelab, add a couple of CD track markers (including a start marker, please) and then Process > Auto Split. See if I’m right.
I have also seen Wavelab insert the file name as the marker name while the user is editing the marker. I’m not sure what combination of keystrokes does this, because I’ve never done it intentionally. But, when auto-splitting, the same naming collision would occur.
Since Wavelab is generating the conflicting file (or marker) name, perhaps it would be better if Wavelab added a suffix like -# or -auto_split_# – replacing the # with 1, then 2, 3, etc. – when this situation arises?