(Moved from clib/cclib.github.io#9)

Currently the script coverage.py considerd an attribute "covered" for a program when at least one unit test parses it. It would be better to tighten the requirement by checking if the unit test actually tests the attribute (uses it in at least one assertion statement).