Improve common integration test infrastructure
Our integration tests are written in a way that they can run just about any script as a testing tools. The base of the test is usually some shell script calling a python script doing the heavy lifting of sending requests and verifying results. These scripts are copy-pasted and could really use some form of unification.
One reason for that can be seen in changes in !3 (merged), where I had to change every single script after a change of a single constant.