Drop use of pifpaf and the "db" pytest mark

We've been using pytest-postgresql for... a year (4117d5ab).

Test Plan

tox -r -e py37

---

Migrated from D3213 (view on Phabricator)

swh/scheduler/tests/test_celery_tasks.py

@@ -50,7 +50,6 @@ def test_multiping(swh_app, celery_session_worker):
         assert ("OK (kw={'i': %s})" % i) in results
 
 
-@pytest.mark.db
 def test_scheduler_fixture(swh_app, celery_session_worker, swh_scheduler):
     "Test that the scheduler fixture works properly"
     task_type = swh_scheduler.get_task_type("swh-test-ping")
@@ -67,7 +66,6 @@ def test_scheduler_fixture(swh_app, celery_session_worker, swh_scheduler):
         AsyncResult(id=task["backend_id"]).get()
 
 
-@pytest.mark.db
 def test_task_return_value(swh_app, celery_session_worker, swh_scheduler):
     task_type = swh_scheduler.get_task_type("swh-test-add")
     assert task_type
@@ -82,7 +80,6 @@ def test_task_return_value(swh_app, celery_session_worker, swh_scheduler):
     assert value == 42
 
 
-@pytest.mark.db
 def test_task_exception(swh_app, celery_session_worker, swh_scheduler):
     task_type = swh_scheduler.get_task_type("swh-test-error")
     assert task_type