I don't think whether they're driving or not is a meaningful distinction at this level

But it does! For example, if the driver seat offers better protection than the rest of the car, and women are more often than men in one of the other seats, it would explain the results and the dummy doesn't add much.

But if the fatality rate for women in the front passenger seat, for example, is the same as for men in that same seat, that's were probably having an "anatomically correct female crash test dummy" can be very helpful in understanding why these crashes are killing more women than men.

so I'm making the fairly safe assumption they put dummies in various different seating arrangements

The source doesn't use data from crash test dummies but from real life crashes. So we can't take seating arrangements for granted if it could meaningfully effect the numbers.

The stats apparently originate from the US government, so it's going to be a pretty big sample size that should average out any differences in seating position.

The sample size is irrelevant if cultural factors exist that could skew the results. Cultural factors like men are more commonly taxi/Uber/bus drivers, men are more likely to drive with their partner as a passenger than the inverse, etc.

I don't think there are really any conclusive after stats as the product was only introduced to the market a couple of years ago

That's a fair point, I don't expect there would be enough data for anything conclusive.

there's a correlation with male safety improving with advances in the crash test process that aren't reflected equivalently with women's safety

That would be an interesting read. I'll have a look for it.