Two, the definitions are all over the map - some 6 years, most 7, the SF study 10 years. Some specify no anti-virals for the non-progressor definition, without saying whether the comparison groups are or aren't also no anti-viral classes. One study finds strong evidence that strains of HIV exist that do not cause disease, probably as a special case but also a monkey wrench thrown into non-progessor statistics.
Most of the rest of the studies there manage to show that defined non progressors really are healthier, not just taking longer, which is a rather limited statement that should surprise no one (since they are selected-defined as HIV+ for long periods with good blood scores and no AIDs).
Obviously, the studies are primarily designed to address guesses about how it is working or what subsystem is involved. They are not primarily directed at mapping the epidemology (how many go from this to that category in this or that long, etc). The result is less than a clear picture, but certainly I see nothing so far to support the "1-2%, corrected for everything" figure.