Surely more productive industrial researchers are generally paid more. Many firms even give explicit bonuses on a per patent basis.

Yes, but the bonuses I've heard of are in the hundreds to thousands of dollars range, at companies committed to patenting like IBM. This isn't going to make a big difference to lifetime incomes where the range is 1-3 million dollars although the data may be rich enough to spot these effects (and how many patents is even '4x'? 4 patents on average per person?), and I suspect these bonuses come at the expense of salaries & benefits. (I know that's how I'd regard it as a manager: shifting risk from the company to the employee.)

And I think you're forgetting that income did increase with each standard deviation by an amount somewhat comparable to my suggested numbers for patents, so we're not explaining why IQ did not increase income whatsoever, but why it increased it relatively little, why the patenters apparently captured relatively little of the value.