This means test files are imported as modules, not run as scripts. THey
can still be run individually if so desired. Test coverage is measured,
and an html report generated in htmlcov/. Test cases are automativally
discovered, so they need not be listed twice. An overall
passed/failed/skipped summary is printed, rather than for each file.
Arguments passed to ./test are passed on to unit2. For instance
./test -v
will print each test name as it is run.
We use test.sh to consolidate test cases in one file so the README
would not be outdated easier rather than the hard-coded test commands
in a code block.