Remove redundant db indexes

[coalest]

Doesn't resolve any issue.

Description

While familiarizing myself with the database schema, I noticed we have some indexes that are redundant.

Context: Composite indexes are able to be used when query conditions include fields involving the leftmost fields. So an index on (a, b) can be used for querying by a and b or by a alone. (But not queries only involving b). We currently have a few indexs on both (a, b) and (a), so this PR is to remove the indexes on (a) leaving only the composite index.

Why remove these redundant indexes? It's not a priority but they do take up storage space in the database and they have a performance impact on write operations (because whenever those columns are updated, the indexes also need to be updated).

Type of change

Technical debt?

How Has This Been Tested?

See screenshots below.

Screenshots

Example before and after for removing the index on user_id from the user_roles table

Before:
You can see it is using the index only on user_id (index_users_roles_on_user_id).

After:
That index no longer exists after the migration, but it is still able to perform an index scan by using the compound index on user_id, role_id.

Note: If you want to test this locally, you might have to insert rows into the database because when the table is too small postgres sometimes chooses to use a sequential scan.

[dorner]

Interesting - the users_roles table was created via a generator from Rolify. But yeah, they do seem redundant. Thanks!