On the question of one's ability to emphasize with people who don't look like them...I'm sorry, I don't buy that (fully).
I've added the caveat of "fully," because there was research done awhile back that I read about (forget where/when) that confirmed that yes, children do find it easier to emphasize with people who look like them, at least to a point. But leaving aside personal feelings on the matter, if there's a linear relationship between one's physical appearance and the type of material they're into, I'm not seeing it on the ground. I work across libraries in a variety of different locations with a variety of different backgrounds, mainly outside the Anglosphere. I've got a good sense of what "kids these days" are into, and it remains the same across locations. If there was a linear relationship between one's culture/appearance and their tastes in material, there should be some variation, but there isn't. Stuff like Harry Potter is just as popular in an area with a high level of Arabic & Chinese expats, as another area with Indian & South Korean expats, as another area where Euro-Australians are the main group. And that remains the same for all kids' series. I can answer anyone at any location what a popular kid's series is, and the answer will remain the same regardless as to what that location is.
I'll accept that what is presented by anecdote can be dismissed by anecdote (hence why I'm not bringing in personal anecdotes), but while there's definitely some relationship between appearance and ability to be invested (per the research I mentioned earlier), I'm skeptical that it's 1:1.
As for the question of historical over-representation, okay, sure, but how does that work retroactively? If you argue that it does, then you have to define how far you want to go back, cross-reference that with the presence of the group in question in the present with the question of the group in the past, and then do the math. Furthermore, you have to question whether you put it on the national level or global level. And if it's on the global level, things get even more complicated. And if you come at a figure, then what? And does this apply to real-world settings, or fictional settings? If we're going back to Overwatch, then how does one account for the presence of non-human characters, and of the human characters following fictional religions (e.g. Ana and Pharah) when it comes to balancing things out? Should Overwatch represent the demographics of today, or the projected demographics of the 2070s?
Don't get me wrong, I'm not an SQW - in case I haven't made it clear in this thread, I despise SQWs, probably even more than SJWs, because they tend to be more aggressive and more toxic. It was SQWs who lost their minds with the casting of Hermione in Cursed Child, SQWs who lost their minds with Tracer, and SQWs who lost their minds with Star Wars (see Rose for instance). But on the flipside, when people bring in percentages, I don't think there's a way that argument can be won by anyone because everyone's going to use different starting points and assumptions, and there's an infinite no. of traits you can consider.
So on one hand, I have no time for people who whine about "forced diversity," because almost all the time, that isn't actually a thing. On the other, if people say "setting X needs more of group Y," then they'd need an actual argument and be able to quantify it.