Duplicate entries created in Merge table

[samuelbray32]

Describe the bug

• Found that some of the entries in SpikeSortingOutput.CuratedSpikeSorting are duplicated with different merge_ids. This should be prevented by the UUID hashing in Merge._merge_insert().
• Possible that these were due to a change in how we hash Fault-permit insert and remove mutual exclusivity protections on Merge #824
  • If so, should _merge_insert check for a matching entry in the part table for a given source primary key before generating a UUID and inserting? This would help prevent overlapping entries going forward in case of other hash method changes
  • Side note: Essentially this is the inverse problem of Merge key conflict in PositionOutput #915

To Reproduce
Example duplicate key:

part_key = {
    "curation_id": 1,
    "nwb_file_name": "Winnie20220717_.nwb",
    "sort_group_id": 13,
    "sort_interval_name": "12_lineartrack",
    "preproc_params_name": "franklab_tetrode_hippocampus",
    "team_name": "ms_stim",
    "sorter": "mountainsort4",
    "sorter_params_name": "franklab_tetrode_hippocampus_30KHz_tmp",
    "artifact_removed_interval_list_name": "Winnie20220717_.nwb_12_lineartrack_13_franklab_tetrode_hippocampus_ampl_2000_prop_75_artifact_removed_valid_times",
}

SpikeSortingOutput().CuratedSpikeSorting() & part_key

[samuelbray32]

Can confirm the difference is due to inclusion of the source table in the hash generation. duplicate entry UUIDs match hash results with and without the source table respectively.

Solution: Implement check for existing key in part table prior to UUID generation and insert