Optimize skia rendering loop

[tushar5526]

Describe the bug
glfw uses double buffers while rendering graphics, which means re-drawing the frame from scratch every time. The problem is that we don't have/store the commands previously executed in draw, setup or event_handlers. We get the new command from draw to execute on each new frame.

The current hacky fix was to, take a snapshot of the frontbuffer - draw the new commands on top of it and then switch buffers. We repeat this process every frame.

This issue deals with exploring ways, how to make this work more efficiently.

Refer to this section in code: https://github.com/p5py/p5/blob/master/p5/sketch/Skia2DRenderer/base.py#L127-L131