test: two background_tasks wait_for timeouts flaked in CI on PR #864 #865

Seen on PR #864 (test/docs-only changes). Each failed once in CI, then passed on a re-run. Neither reproduced locally: 5 runs of the outbox file, 3 runs of the invalidation and outbox folders against Redis, and 2 runs of the whole background_tasks suite under xdist with 4 workers and Redis.

• Attempt 1: modules/background_tasks/tests/outbox/delivery/test_background_tasks_outbox_eager_platform.py::test_eager_publish_from_a_plain_session_is_delivered
AssertionError: wait_for(delivered) not satisfied within 15.0s; last error: TimeoutError()
• Attempt 2: modules/background_tasks/tests/core/invalidation/test_background_tasks_invalidation_redis.py::TestOriginFilterAcrossTheWire::test_a_second_worker_does_apply_it
wait_for(<lambda>) not satisfied within 5.0s, preceded by Invalidation broadcast failed for channel 'qa.invalidation.2788' (the publish to the CI Redis service failed)

Run: https://github.com/antosubash/simple_module_python/actions/runs/42202400408 (attempts 1 and 2)

Both tests wait on asynchronous delivery with a fixed deadline, so this looks like CI load or Redis-service latency. Options: give the wait a longer deadline in CI, or retry the Redis publish in the test harness.

https://claude.ai/code/session_01FXHA3ZicrtZ4C5QPEgxfgq