Metadata visualization shows table key not relationship key for relationships

[frances-h]

Environment Details

Please indicate the following details about the environment in which you found the bug:

• SDV version:
• Python version:
• Operating System:

Error Description

Currently, Metadata visualization always shows the parent table primary key in the relationship, even if the 'parent_primary_key' value in the relationship is a different column. The metadata visualization should show the values set by the relationship to help with visual debugging when validation fails.

Steps to reproduce

metadata = Metadata.load_from_dict({
    'tables': {
        'parent': {
            'columns': {
                'pk': {'sdtype': 'id'},
                'other': {'sdtype': 'categorical'},
            },
            'primary_key': 'pk',
        },
        'child': {
            'columns': {
                'fk': {'sdtype': 'id'},
            },
        },
    },
    'relationships': [{
        'parent_table_name': 'parent',
        'child_table_name': 'child',
        'parent_primary_key': 'other',
        'child_foreign_key': 'fk',
    }]
})
metadata.visualize()

[npatki]

Expected Behavior

In the above case, we expect that the edge should be listed as fk --> other instead of what it currently is labeled (fk -- pk). This edge should be based on what the 'relationships' section of the metadata has.

Note that within the parent table, the primary key should remain as pk. That should be based on what the 'tables' part of the metadata has.

So we expect that the visualization should showcase the discrepancy (the edge would be marked differently than the primary key in the table).