Tests: Speed up MR and deployment testing

At the moment full deployment testing takes over an hour, possibly around two hours, and even regular testing takes quite a while. This slows down development considerably. Find for ways to speed it up. Options include:

* Parallelize testing. Allow tests using independent singleton fixtures to execute in parallel. It's not clear how to do that, as pytest doesn't seem to support it.
* Remove less important tests, or relegate them to scheduled testing only.
* Figure out how to only trigger tests that are relevant to the changes
* Figure out how to speed up particularly-slow tests
* Figure out how to speed up deployment. E.g. deploy some things in parallel, or get rid of the slow-building jq dependency in cloud deployments.

Start by measuring the pytest-based regular and full deployment test runs, including the particularly slow tests and cloud deployment steps. Compare parallelization, selective testing, scheduled tests, and the jq dependency as possible directions; done means a documented, measurable reduction in testing or deployment time without losing required coverage.

Written by the indexing model from the issue text.