Disease, for example, is mostly a product of settled societies, not hunter-gatherers. Crime was a different issue as well, when tribes knew all the members. And sure, tribes had conflicts, but they were often not as violent as the total wars of the last couple centuries.
I'm surprised so many people who know nothing except modern society are so quick to make sweeping assumptions about a distant and unrecorded past.
Same logic applies to medicine and insurance.
https://evolutionistx.wordpress.com/2015/06/08/no-hunter-gat...
The link also mischaracterises the argument that anthropologists make: they do not believe that foragers had a fundamentally different ("noble savage in a Marxist feminist garden") character than moderns, but that their incentives and constraints differed. It is hard to find people to war with if you are living at extremely low population density, or to carry out ambush raids along long sightlines in the savannah. You are probably less likely to murder your neighbours' children if they are your nieces, or fail to share food with your brother's wife. Hunter-gatherer social technology exploiting human biases (e.g. toward kin) is often ignored in favour of wishful retrocasting of our own political viewpoints.
Globally, foragers may not have enjoyed the low levels of violence we do in modern society, but they may have been competitive with gardening/agricultural/tribal/state societies for millennia.
As for relatedness, that's relative (no pun intended). If everyone within 200 square kilometres is at a genetic distance of a second or third cousin, then a second cousin is the base for a non-relative. In any case, regardless of why precisely it is that violent death is so common pre-modern and especially pre-state societies, it's abundantly clear that it is.
Regarding low population densities, you'll notice that Australian Aborigine tribes are represented in that chart and exhibit murder rates that are far higher than those in modern societies. They do however have the lowest homicide rates among hunter gatherer tribes. The highest homicide rates are found in the Amazon rainforest, where population densities would be highest. So it appears that the two factors that contribute to high homicide rates are pre-state/agricultural society and high population density, and that in the case of the tribes living in the most sparsely populated regions, the inhibiting effect of extremely low population densities on homicide rates does not make up for aggravating effect of living in pre-state societies.
Of course relatedness is relative, but the argument is that encounter rates of whatever a group deemed "non-relative" were relatively low and that hostile contact was very costly/relatively unproductive. The "other" was known, cementing feelings of kinship, but distant.
The most violent societies according to your charts are post-band but pre-state with relatively high population density. That is not in contradiction with the "foragers nonviolent compared to other pre-state and comparable to earlier state societies" model.
The interesting question is whether there may have been two optima for human organisation - one at very low population density mobile foraging, the other at the point of the modern state. The modern state may be preferable to the first optimum, but the idea that it takes 10,000 of evolutionary search time to find the second hill is fascinating.
This is also not a settled question. Many scientists believe that the incentives and constraints on settled peoples led to very different genetic pressures, thus leading to a very different character in modern societies than those from from a mere 10k-20k years ago.
how is this a thing? Maybe you're thinking of large scale epidemics, but the most common cause of death in tribal settings are (modernly) treatable infections.