Andrew Lyons authored Oct 15, 2015 and Damyon Wiese committed Oct 15, 2015

The benefit is that this becomes one test per external function instead of one
test testing them all - so more information is shown on failure and we get lots
more dots (we love dots!).