Feature request: Implement scale_type "original" / scale_type "passX" for shader presets

Description

When developing shaders, it's often faster to have pre-processing passes that output a certain resolution (the resolution of your LUT, for example). There's the scale_type "absolute" for this purpose.

However, for passes that follow the pre-processing passes, there is no adequate way to control output size.
Scale_type "source" is clearly not what is wanted: The LUT size is often completely independent of actual content being rendered.
Scale_type "viewport" will only allow the passes rendering at exactly the output resolution to use pre-processed passes sensibly, as they can have output sizes tied to the viewport.
Scale_type "absolute" is not flexible enough to solve this use case.

Expected behavior

I would like to have more control over output sizes, namely one of the following:

• Implement scale_type "original", which refers to the input size. This allows all N pre-processing passes to happen at the very top, with the N+1th pass to use scale_type "original". This gives complete freedom but enforces a certain structure.
• Implement scale_type "passN" (or "alias ") which would scale relatively to either that pass index, or the aliased pass with that name. This would give complete freedom, and pre-processing shaders could be woven into the pipeline at arbitrary stages.

Version/Commit

Feature not yet implemented as of tag v1.10.3

Environment information

Irrelevant as this is a feature request.

The issue identifies no source files, tests, or entry points; first locate the shader preset scale-type handling and existing absolute, source, and viewport behavior. Before coding, clarify whether completion means supporting original, passN, or a named alias, then verify that the selected form controls downstream pass output sizes as described.