Debug current generate issues
Fix current generate issues #19 (closed) and #46 (closed)
For #19 (closed) and duplicate sample in csv file this could be due to many cause:
- the wrong dataframe was used when saving csv sample for pointer i. (s should be use in place of strip)
- no check was made to not write an existing sampling from a previous generation.
So to avoid duplicate sample this merge request :
- fix the dataframe to be used when saving csv
- add code to check if a sample already exist in train/val/test csv files
- add a supplementary no_duplicate in save dataframe.