Each frame can be set a different position, width and height, allowing to update only a part of the GIF.

The each are encoded in the imageDescriptor and should use lsb() like width and height already does.

We may need to add an options parameters to the addFrame function.

const encoder = new GIF(width, height);

// Bad, too much arguments
encoder.addFrame(context, delay, x, y, width, height);
// Better ?
encoder.addFrame(context, delay, {
	position: [x, y],
	size: [width, height],
});

As of now, we extract data from the whole context. This new position option can be mistaken for the position from where to extract data.
This is the same issue as drawImage from the CanvasRenderingContext2D API.