feat: add Group Community TA audience for cohorted content_reported notifications

[pavanhalesh]

Description

Adds a new notification audience filter to include Group Community TA / Group Moderator users for cohorted content_reported notifications.

Updates the discussion notification payload to include the cohort id so notifications can target the correct Group TAs.

How it works

When content is reported from a cohorted discussion thread, we pass cohort_for_group_tas=[<cohort_id>].

The new GroupTAinCohortFilter returns only users who:

belong to that cohort, and

have the Group Moderator role for that course.

Testing

Verified backend logic via Django shell by creating Cohort A/B, creating users, assigning Group Moderator role, and validating:

Cohort A → group_ta_a

Cohort B → group_ta_b

Removing TA from cohort → returns empty list for that cohort

End-to-end UI validation is currently blocked locally due to the Discussion/MongoDB connectivity issue in devstack.

Notes

No DB migrations.

@asharma4-sonata Updated branch/PR/commit naming to match conventions (no internal IDs in titles/messages).
Raised PR from fork targeting edx/edx-platform:release/ulmo using branch feature/group-community-ta-content-reported-audience.

Please review.

[asharma4-sonata]

I will work on it tomorrow.

[asharma4-sonata]

We can use this as subquery for group_moderator_user_ids_qs below.
This will ensure DB is not hit twice -> will lower the latency.
Passing the QuerySet directly into the next filter allows the database to handle the join internally.

[pavanhalesh]

I’ve updated this to pass the cohort users queryset directly into the Role filter so the database performs the join internally, avoiding an extra DB hit and improving latency.

[asharma4-sonata]

Not required. what if a large course contain thousands of users. We don't want to evaluate the QS in Python memory just to pass it as IN clause to hit database packet size limits.

[pavanhalesh]

Makes sense — thanks for calling this out.
I’ve removed the Python list evaluation and now keep this as a queryset to avoid memory overhead and large IN clauses.

[asharma4-sonata]

should include .distinct(). If a user is associated with multiple roles or groups in a way that creates duplicate rows.

[pavanhalesh]

I’ve added .distinct() to ensure we don’t return duplicate user IDs when users are associated via multiple roles or relationships.

[asharma4-sonata]

Hi @pavanhalesh, the logic here correctly identifies the Group Moderators for specific cohorts. However, I have a few concerns regarding memory safety and query efficiency at scale, which I have detailed in the inline comments.

[pavanhalesh]

Thanks for the review and detailed guidance, Amal.
I’ve updated the implementation to be memory-safe and more efficient at scale by using subqueries, avoiding queryset evaluation in Python, and adding .distinct().
Changes are pushed to the same PR — please let me know if you’d like any further refinements.