fix test_filters to use set to compare dag id's in dag run tests

[rawwar]

CI failed for test_filters in test_dag_run.py(fastapi).

This PR fixes it

Issue. Test should get all dag runs with "success" state. But, the order of these dag run's isn't relevant to this filter test. Hence, using a set instead of list.

EDIT: Looks like this is an issue across tests for list dag_run. I'll update this PR to fix all of its tests.

[pierrejeambrun]

I do not agree with that. Ordering of items returned by the API should be deterministic. Fixing the flakiness like this does not actually solve the problem I belive.

[potiuk]

I do not agree with that. Ordering of items returned by the API should be deterministic. Fixing the flakiness like this does not actually solve the problem I belive.

I agree it's a better solution

[uranusjr]

I don’t know about the new implementation, but in the Connextion implementation, the result is actually sorted, but it’s against a very weird default (id) that isn’t really deterministic in the tests because when you insert things through the ORM it’s not guaranteed the id would be the same as the order you insert them (due to both database and ORM implementation). So the tests will need fixing in any case unless we change the default.