>>5
But does this actually work? I don't know Zig well enough to judge whether this would type check (or... whatever) in the macro facility.
If you demonstrated a simple proof of concept that can evaluate arbitrary subexpressions in place of a word/atom, that would be a good starting point.