FIX: Refactor prepare_name_pairs_pd to pass arguments to create_positive_negative_samples

[chrispyl]

prepare_name_pairs_pd is not passing the column names for uid, correct, and positive set columns to create_positive_negative_samples.

In this pr:

• correct_col, uid_col, gt_uid_col, preprocessed_col, gt_preprocessed_col become parameters for prepare_name_pairs_pd
• the corresponding hardcoded values of prepare_name_pairs_pd become default values for these columns and are replaced by the variables
• the above columns are passed from create_training_name_pairs to prepare_name_pairs for both pandas and spark versions
• uid_col, correct_col, positive_set_col are passed from prepare_name_pairs_pd to create_positive_negative_samples