Fix flakey bart robot unit test

[olliesilvester]

One of our tests for the bart robot occasionally fails. We should fix this

tests/devices/unit_tests/test_bart_robot.py::test_given_program_not_running_and_pin_unmounts_then_mounts_when_load_pin_then_pin_loaded - AssertionError: Expected 'trigger' to have been called once. Called 0 times.

Or see https://github.com/DiamondLightSource/dodal/actions/runs/12165707719/job/33930319285

Acceptance Criteria

• Test made more robust

[paulorochaoliveira]

FYI, I'm taking a look at this flaky test right now.

[paulorochaoliveira]

It seems like what makes this test flaky is the sleep time sometimes being to short. Increasing the sleep time would most likely fix the issue.
Another more robust solution is to replace arbitrary sleeps with an event-driven approach. For example, use asyncio.Event to synchronize the test with the actual behavior.
What do you think @olliesilvester ?

[olliesilvester]

I think increasing the sleep is good enough for this issue, however I like the idea of using asyncio.Event instead. I will make an issue to replace all of our arbitrary sleeps in tests instead of this! Thanks!

[paulorochaoliveira]

Should I open a PR to at least close this issue?

[olliesilvester]

Yes, go ahead, thank you

[olliesilvester]

This came up again

[DominicOram]

I think we need to properly mock it and rethink how the test works rather than just change the length of sleeps. e.g. we could patch sleep like in

dodal/tests/devices/unit_tests/test_thawer.py

Line 29 in f93183f

async def test_when_thawing_triggered_then_turn_on_sleep_and_turn_off(

so that we can be very specific on when things should be happening