This PR aims to make it possible to optimize most of the effects which can be rendered with a single fragment shader pass. The old pipeline uses 3 passes:

• Compute pass for the effect
• Compute pass for the overlay engine
• Fragment shader pass for rendering

The new render code combines all three together in a single fragment shader pass, allowing us to also make efficient use of GL scissor. By reducing the passes, we also reduce the CPU usage.

The idea of how the shaders work:

• All of them share the same headers and the same uniforms, width, height, current_time
• Each effect needs to provide a function, vec4 effect_color(vec2 pos) which gives the color from the effect.
• Each overlay engine provides a function, vec4 overlay_effect(vec2 pos). It may call effect_color() to get the color of the effect, or set a color on its own. For example the none overlay effect simply returns effect_color(pos).
• The main function is the same for all combinations, it just calls overlay_effect(pos).

The headers, effect, overlay and main function code are simply concatenated together to form a single shader which is then executed.