Source (the Working Copy) CSV files.
When CSV files is set as a data source, the SFDMU analyses and tries to fix the raw source CSV files you've provided, then it copies final versions of them into the special directory. These files become the working copy of the raw CSV sources, which will be actually used to import data into the target org.
You can review these files and see how the SFDMU really handles your CSVs.
The working versions of the source CSV files can be found in the ./source subdirectory.
Each file has added postfix of _source:
If multiple Object Sets are used, the first Object Set source is always placed in the main ./source subdirectory, and the sources of the rest Object Sets starting from the Object Set #2 are being placed into ./source/object-set-<ObjectSet-index>: