Rakeness and beyond in zero-complexity digital compressed sensing: A down-to-bits case study