I use that method all the time, very handy. It would be handy to have a convert alpha to scalar as well, that would extend the use of an RGB file. But I don't really know if the file would be larger with an alpha channel than without. If I need 4 masks, I just use an extra grayscale file or even bitmap.
Making them is easy. What I usually do is make a (base) layered PS file (map on bottom, all necessary masking on top of that), then copy the layers into the appropriate channels of the final RGB masking file(s).