Shape tests

[davepagurek]

This adds unit tests for the functionality within beginShape/endShape in 2D and WebGL mode so that we can use this as a checklist when refactoring the internals for shape drawing.

Some notes:

• This doesn't test combinations of bezier + quadratic vertices within the same shape, since that isn't currently supported
• There seems to be a bug where curveTightness is treated differently within WebGL and 2D modes. I'm leaving the test cases in for now, but we should expect one or the other to break in the future when we use a common implementation.

cc @GregStanton

[GregStanton]

Thanks @davepagurek! I just remembered that I previously made some informal test cases for a couple types of composite curves that aren't currently supported, when I was testing my proof of concept for the refactoring project. They cover mixing quadratic and cubic Béziers, as well as line segments and Catmull-Rom splines. I could add tests based on those later, if that sounds good.

[Qianqianye]

Thanks @davepagurek @GregStanton. Is this ready to be merged?

[GregStanton]

Should we add tests for POINTS, LINES, TRIANGLES, and QUADS?

It looks like TESS isn't covered here either, but I seem to remember a discussion about this feature no longer being necessary; it's currently used in the reference for beginShape() in the last two examples, but replacing beginShape(TESS) with beginShape() has no apparent effect. If it's no longer necessary, I'll make a reminder to remove it from the reference once we update it for p5.js 2.0.

[davepagurek]

TESS is the explicit name for the default shape mode, so testing beginShape() with no arguments covers the same thing, so I think we're good there.

Adding POINTS, LINES, TRIANGLES, and QUADSwould be a good idea, I can add those!

[GregStanton]

Great! Thanks @davepagurek!

About TESS, it looks like that's the explicit name for the default shape kind in 3D mode, but as you may know, there is no explicit name for the default shape kind in 2D mode (it's just set to null). The reference doesn't specify the default in either case, and the explanation for TESS seems unclear. I'll address this as part of the refactoring, in accordance with #6679. We can discuss the details elsewhere, now that I know it's not relevant to this PR.