[BUG] Fix Test Model method

**Describe the bug**
Currently, the `test_dataloader` object is being used for testing **and** validation.
We should further discuss if it's required that we return all label predictions

**To Reproduce**
Run experiment

**Expected behavior**
Different dataset for validation and test

**Screenshots**
N/R

**Hardware**
N/R

**Additional context**
N/R

Start by running the experiment and locating where the `test_dataloader` object is used for both validation and testing. Check how label predictions are returned and determine the intended separate validation and test datasets. Done means validation and test use different datasets, with the prediction behavior clarified.

Written by the indexing model from the issue text.