Add CI job to run system tests

[callumforrester]

Fixes #630

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.92%. Comparing base (86d9470) to head (7071fed).
Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #761   +/-   ##
=======================================
  Coverage   92.92%   92.92%           
=======================================
  Files          37       37           
  Lines        2063     2063           
=======================================
  Hits         1917     1917           
  Misses        146      146           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ZohebShaikh]

Looks good !

[ZohebShaikh]

REQUIRES_AUTH = bool(os.environ.get("REQUIRES_AUTH", 0))
REQUIRES_AUTH_MESSAGE = """
Authentication credentials are required to run this test.
The test has been skipped because authentication is currently disabled.
For more details, see: https://github.com/DiamondLightSource/blueapi/issues/676.
To enable and execute these tests, set `REQUIRES_AUTH=1` and provide valid credentials.
"""
requires_auth = pytest.mark.skipif(not REQUIRES_AUTH, reason=REQUIRES_AUTH_MESSAGE)

and then decorated with

@requires_auth

This will help us to run the tests on the local system even if we are skipping them on the CI

[callumforrester]

I like the longer message, will implement.

I'm not a fan of having special steps to make the tests work, under this you would have to somehow know that you need to do REQUIRES_AUTH=False tox -e system-tests. A dev environment that works in a predictable way out-of-box is better IMO. I'm also not hugely in favour of having the local tests work differently to the CI, we were in that situation before we made bluesky-stomp and it caused lots of debugging problems. I would rather mark as xfail, clearly showing we know there's a problem, and fix it when we can. On a practical level it makes no difference, following the instructions for the system tests locally will result in a success message.

[DiamondJoseph]

Suggested change

	attempts_remaining = 20
	while attempts_remaining > 0:
	try:
	client.get_environment()
	return
	except ConnectionError:
	...
	attempts_remaining -= 1
	time.sleep(0.5)
	raise TimeoutError("No connection to the blueapi server")
	for _ in range(20):
	try:
	client.get_environment()
	return
	except ConnectionError:
	...
	time.sleep(0.5)
	raise TimeoutError("No connection to the blueapi server")

nit: Why keep track when Python'll do it for ya mate

[callumforrester]

Good shout, see also #763

[DiamondJoseph]

No longer running tests against ActiveMQ? Seems reasonable to me, given that the image is so deprecated, but we could replace it with Apache Artemis, just to make sure our Stomp interactions are generally valid.

[callumforrester]

We run the system tests of https://github.com/DiamondLightSource/bluesky-stomp/ against activemq and rabbitmq, which I think is sufficient