run-tests: introduce --random for running tests in random error This is convenient when investigating failures one by one to avoid running the same fixed tests first every time. [ original upstream message ]

(binary:application/octet-stream)