Tests time out on Travis

[LourensVeen]

The tests are very slow, and even slower on Travis, where they have now started to time out regularly. This should be improved.

• Add measurement and get the timings of a test run. How long does it take to start and stop the containers? When does each test start and stop? Are things running in parallel or not? Output ideally to the console, so that we can test on Travis as well.
• Try to speed up the tests, possibly by manually parallelising the scheduler tests?
• Travis times out after not receiving any console output for 10 minutes. We have all the scheduler tests in a single file, which puts them on a single line, and it seems that there's a line buffer somewhere. Can we remove that buffer, or split the tests over multiple files so that Travis sees output more frequently?

[LourensVeen]

We're not using Travis anymore, and the tests are running fine on GitHub Actions. They still take quite a while to run, about 15 minutes. Locally it's the same, which does slow down development.

The main things that take time seem to be building the test container, installing software (many versions of Python, dependencies for each, etc.), and then waiting for slurm to pick up and run jobs.

Some ideas:

• We're using docker-compose to build the test image, and I think that uses the old fashioned build rather than buildx. Maybe we can build it separately, or maybe a newer docker-compose (we're still using the now-unsupported ancient Python version) will use buildx?
• We're currently using 8 threads for the scheduler tests. CPU load isn't very high though on my laptop, because we're mostly waiting around for jobs to get picked up or results accounted. So maybe we can just increase the number of parallel tests greatly here?