I was discussing the period when we humans evolved. Not so large groups then. Evolutionary pressure creating psychopaths from a group evolutionary basis seems unlikely (these generally need quite specific circumstances to work).

(Your argument also miss that psychopaths manipulate and damage the direct environment around them most.)

Perhaps psychopathy is like being bald - no real advantages, perhaps a small disadvantage, but it doesn't disadvantage you enough to stop you having a few kids so the genes get passed on anyway.

Indeed, if psychopathy is a combination of traits ('feels no empathy' and 'good at faking empathy') perhaps the survivors had a combination of traits that helped them to survive, while those with one or the other failed to survive.

Uh, no...

There are large effects on behavior of psychopathy, people that are affected by a psychopath will mostly -- at a minimum -- make a point of warning everyone else they know. That is very detrimental for the psychopath. (Note that how we communicate really applies evolutionary pressure. Just check the number of muscles in a human face compared to most anything else. This is not a small effect.)

Also, psychopathy seems to be partly environmental and partly from genetics. And exist in most populations.

All this suggests an evolved behavioral strategy.

Researchers need to check for your hypothesis (the large different effects almost cancels each others out in many types of historic societies(!), so it is just a random genetic drift) since afaik it is a standard hypothesis, but it does seem very unlikely.