Dataset merge using filename index introduces redundant columns

[JuliaDima]

Issue

Problem Description

The _get_combined_labeled_data method in session.py merges two dataframes on filename as index and left as merge type. This creates new column (usually label_x and label_y as the label column appears in both datasets). When using the dataset further, one can encounter errors due to improper label column.

Expected Behaviour

The merge should yield a single label column, taking into account what values to choose from first or second dataset (i.e., in case there is an update for a filename, take the values from the active learning dataset since this suggests updated labels for a cutout).

How Can It Be Tested or Reproduced

Labelling some files using the AM GUI, saving the labels, then using the labels dataset.