1. The unit_of_measure column is currently part of the tally_snapshot uniqueness constraint, but it is no longer needed and can be removed from the table. The issue arises because PostgreSQL does not treat NULL values as part of a uniqueness constraint, which can result in duplicate records.
2. The columns snapshot_date, org_id, product_id, and granularity currently allow NULL values, but they should have NOT NULL constraints to ensure data integrity.

We discovered an issue in CI pipelines which we suspect there might be a "potential" race condition during the retally(same tally rerun again) process, where the tally_state table is not being properly locked. This could lead to the creation of duplicate entries in the tally_snapshot table.

TODO - split cards

• API to delete rows where there are nulls in any of those columns
• THEN Add constraint (liquibase)
• DROP UOM column (liquibase) from tally_snapshot, adjust the the uniqueness constraint if it doesn't update that automatically
•  

Acceptance criteria:

1. unit_of_measure column is removed from the table.
2. add billing_provider and billing_account_id columns into the unique index.
3. Not null constraints are added to the org_id, snapshot_date, granularity, and product_id columns.
4. Enable Test MR: SWATCH-3466 [Make the sync_tally method more robust|iqe test]
5. Make sure in stage applying unique constraint is quick and doesn't cause any db issues.