add Negative Data training and testing

[Ertz012]

Proposed changes

added a feature that trains and tests on negative data this should reduce false positive segmentations on background

Related issues

Necessary checks

• [ ] Update poetry package version semantically
• [ ] Write documentation
• [x] Create issues for future work
• [x] Test on your machine

[Ertz012]

TODO: split negative data in training and test

[jaagut]

Todo: make superclass dataset with params (e g. Size) and collate_fn. Thus, it's defined on every dataset, not just ListDataset