No, it absolutely is not, and your insistence on calling it that shows you don't understand. You know how normal rasterization works, right? There's an idealized scene with 3D objects which are defined mathematically. The purpose of a rendering engine is to determine what luma/chroma values projected to a quantized 2D plane would most faithfully represent the scene.
So the defined position of the object and its faces determines a base pixel value. This is then refined by adding textures, shaders, and lighting. Often, further adjustments are made based on things like the values of adjacent pixels in a buffer.
What does checkerboard rendering do? Well, the defined position of the object and its faces, carrying texture/shading/lighting augmentations from its prior-frame position, determines a base pixel value. Then, further adjustments can be made based on things like the values of adjacent pixels in a buffer.
What does scaling do? The values of adjacent pixels in a buffer determine the final values.
So the technical details of the methods clearly indicate checkerboard is a type of rendering, not a type of scaling. But there's even more reasons that's true:
1. In certain situations, checkerboard can produce results identical to native rendering.
2. Checkerboard can produce luma/chroma values outside the range of surrounding pixels.
3. The inventors and users of the method refer to it as "checkerboard rendering", not any type of upscaling. So does every dev I've seen on GAF.
In short, the case is closed. Checkerboard rendering is just that, and not a type of scaling.