Fix python version used to run tests

[behrtam]

Even if the test runner itself is executed with a specific python version the subprocess used to execute the respective test script always uses the generic 'python' found in the PATH. On most systems that have python2 & 3 installed 'python' is a symlink to 'python2.X'. So this would mean that even if you would run 'python3 ./test/...py' the tests would still be run with python2.

[rootulp]

Good catch! Do you think we could've caught this sooner if we included a DEBUG log line that prints the version of Python that is used to test? Do you think it is worth adding this log line now? I'm torn because the check-exercises.py output is very concise but we explicitly ask contributors to test with both versions and an additional log line might serve to validate that the version they have tested against did in fact work (ex. 2.7 and 3.5).

[behrtam]

I like the idea of adding a log line. If we add one line right before the FAILURE/SUCCESS message I don't think that would ruin the concise output.

$ test-runner  ./test/check-exercises.py hello-world
#  hello-world
......
----------------------------------------------------------------------
Ran 6 tests in 0.000s

OK

TestEnvironment: Python2.7

SUCCESS!

[rootulp]

Looks good to me 👍