Working on new features/optimizations and testing these with a large set of tests, it would be helpful to have a command line flag that allows adding new IR verification rules ("failOn", "counts") to all tests to then check which tests are still passing. The rules of these passing tests can than be strengthened.