Since this test inheriently has `--max-import-depth=1`, by using six, we would
never look at the actual source-code of urllib.parse/urlparse and therefore the
test would never show if we understood the library code good enough that we
could propagate taint out-of-the-box.
All tests moved by one line... that is why the diff is so big
Following the setup I invented for library-tests/taint/unpacking.
TestStep is still a bit annoying, since the output is not easy to eyeball; but
for now I guess we can live with it :)
I honestly didn't get the point of DistinctStringKinds.ql, other than showing we
can handle multiple taint kinds
