More precise testing of plugins

[ericwb]

Instead of relying on a count of the severity and confidence levels found within an example file, make use of Python's native unit testing to verify the results of a plugin.

The existing method of confirming counts can be inaccurate. It's very easy to have a false positive simply because one issue extra was found and one issue was missed, thus giving the same count. It tells nothing of the validation of a particular line of problematic code.

Relates to #352

Signed-off-by: Eric Brown [email protected]