Just to clarify, my particular suggestion is not so much a PixelMath replacement with a more "intuitive" GUI, but a tool that lets you combine several images in a more "visual" style, where you can apply typical PixelMath operations between one image and the one on top. In short, the "layer paradigm" implemented in a simple, yet, I believe, effective way.
Juan, as you point out, the above idea (and an even more detailed "true" layers implementation) adds very little to the PixInsight platform in terms of image processing. But - now referring only to this simplified version of layers - it does provide a more intuitive interface for a not uncommon use of PixelMath operations (not limited to retouching/editing, btw). The reason I mention it is my belief that this shouldn't be hard or too time consuming to write, and possibly not conflicting (much at least, I'd hope) with your vision of quality, usefulness, and high standards for the PixInsight platform.
Better than that, although the UI wouldn't be nearly as fancy as other implementations of layers, it would be close enough to be offered as a "weapon" (if that's in fact a plus), and better yet, even more powerful in some regards, as it allows you to define the math operations that are applied, as opposed to being restricted to a limited number of predefined operations (screen, darken, etc).
On top of that, more advanced features could be added later on, say for example, instead of defining a math operation between two views (layers), also being able to define a process icon as the operation.
It might not be the layers UI many people would like to have, but I'd much rather have the above tool available, than a fancy UI implementation that lacks that functionality. I don't remember the exact look of the screenshots you once posted a while ago, but I remember seeing only predefined operations (screen, lighten, etc), which was somewhat disappointing.
Anyway, just some ideas...
