Tadhg/cogassign no signals

[tadhg-ohiggins]

Try the direct way
And not the indirect way;
Let’s see what happens.

---

Since one of the things I had to do was merge main into the my branch, in order for this PR to be reasonable you probably need to merge main into jmm/resolve-cog-baseline-bug.

I commented on 2483 suggesting something similar to this approach. This removes reliance on signals entirely and instead makes any save() of CognizantAssignment check to see if this is new object creation, and if it is, do the things that were formerly in support.signals.

[JeanMarie-PM]

@tadhg-ohiggins Too many changes to fully check.
But I'm good with moving the code from signals.py to models.py.
I don't know where the call to cog is being made upon submission - hopefully, you've removed sac.sace() from there.
If all tests are passing, you have my approval to merge.

[tadhg-ohiggins]

@JeanMarie-TTS now that main has been merged in, this PR consists of two changed files, one of which is signals.py being deleted entirely.

The change this branch makes is to override the save method of CognizantAssignemt so that whenever it’s invoked, it checks to see if the save is an instance being created or updated. If it’s being created, it then updates the corresponding SingleAuditChecklist appropriately and saves it, then also runs reset_baseline and attempts to update the corresponding dissemination.General entry as well (if one exists).

The logic is intended to be the same as was happening in support/signals.py. With this version, the code is synchronous and the code path more linear.

If the logic in this PR matches what you expect, you can approve it and merge it into jmm/resolve-cog-baseline-bug, and then if the tests on #2483 pass that should be good to merge into main.