As noted in ComPWA/expertsystem#108 and ComPWA/expertsystem#138, test coverage as computed by pytest-cov decreases if running on multiple cores. This causes a mismatch of test coverage computed on CI and computed locally (if run on multiple cores). It would be better if all tests run single core. Question is, how to do this in a DRY way?

Current proposal:

• Implement Set number of cores through global settings module #10
• Use pytest fixture (?) to set default number of threads to 1, for all tests