Improve flexibility of image diff functionality

Add information about the amount of incorrectness in an image, perhaps using color distances.

Consider adding thresholding for "almost-equal" color values that can be passed to the diff functions. This would work well for color diffs but not for location diffs.

I like [mshafae's] idea of looking at the count of incorrect pixels to see how close student work was to some ideal. I think this would work better in cases where error doesn't compound. In recursive fractal tree drawing, for example, more pixels are drawn many layers deep into the recursion which gives the error more time to add up. This gives me some more ideas to add to that automated diff tool I created which can already highlight diffs in red but doesn't provide any other helpful output.

The issue does not name files, tests, or an entry point. Start by locating the image diff functions and reviewing how color and location differences are represented. Define the desired incorrectness measure and threshold behavior before implementing and testing the selected scope.

Written by the indexing model from the issue text.