Introduce test scenes that demonstrate conflation artifacts

[armansito]

Added two scenes that demonstrate the conflation artifact problem as described in #49. The first scene demonstrates adjacent triangles and rects that belong to the same path and use opposite winding.

The second scene demonstrates strokes with overlapping square caps (these strokes are currently expressed as rects painted with the NonZero fill rule).

[DJMcNab]

Scene 1 conflation_artifacts (with feasibly less colourblind-unfriendly colours):

Presumably, all three squares should be solid colour. Note that the second and third squares have an underlaid red #ff0000 background (or well #0000ff blue), whereas the first has no such underlay, so the background black is seeping through.

Scene 2 labyrinth:

I'm not quite sure which part of this is interesting, presumably the extensions at the intersections

[armansito]

@DJMcNab I'd be happy to change the colors for the test case if you prefer that. What you described is correct, in the conflation_artifacts scene, all 3 squares should have a solid color. This demonstrates when opposite winding at adjacent edges "cancel each other out".

The labyrinth scene, if rendered correctly, should not have the "bulging" effect where the rectangles overlap. You can see what it should look like by changing the translation offset of the labyrinth from Affine::translate((20.5, 20.5)) to Affine::translate((20., 20.)) which will cause the edges of the shapes to fall on exact pixel boundaries, leading to 100% coverage of covered pixels and will only show solid lines. Whereas the fractional offsets in the test scene lead to partial pixel coverage, which makes the artifact visible where coverage is over-counted on overlapping regions which affects the alpha mask (I believe in this case you get 100% alpha instead of 50% as with the non-overlapping edges).

[DJMcNab]

My only experience with colourblind-friendliness is from watching Cracking The Cryptic, where they prefer to use their blue and orange colours when colouring.

I strongly suspect that red and green aren't a good set of colours to use, but I'm not certain what a good set would be.

I don't have colourblindness, so without using external tools, I can't say if the scene in this PR is accessible.

[armansito]

I changed the colors to the following. These foreground/background colors should maintain a high contrast across the types of colorblindness I'm aware of based on this tool.

[raphlinus]

Thanks, these are useful for showing conflation artifacts and will be helpful for verifying that attempts to remove them actually work.