Redo the design of functional tests

[ericwb]

Describe the bug
Currently the functional tests just measure a count of the number of high, medium, low confidence and severity for a file. This mostly works, but can hide false positives and isn't the best way to do an absolute test that a function gets a desired result. It relies on counts and not whether an individual line caused an bandit error or warning.

To Reproduce
Look at functional tests and how they work.

Expected behavior
Instead of the current method, I think we need to devise a way that each line is checked rather than comparing a total count for a file. Maybe we don't even need the files. Maybe these just become better unit tests of each plugin.

Bandit version

bandit 1.4.1
python version = 2.7.10 (default, Oct 6 2017, 22:29:07) [GCC 4.2.1 Compatible Apple LLVM 9.0.0 (clang-900.0.31)]