On a related note: if it is true, does that suggest that, as far as we take the diminishing utility of money for granted, by using extrinsic rewards, we are reducing the number of extreme performers? (in so far as we can't keep giving exponential rewards, and money/tokens/what have you motivates in proportion to their utility)

Paying people is complex. Companies often pay their employees often market wages and don't pay to optimize motivation. Even when it comes to optimizing motivation fairness perception matters a lot.

What else would? (like, even if stereotype threat existed as a significant force, it seems far less clear to me how that finding could realistically impact any policies or our behaviors)

If stereotype threat is a force that exists there are likely variables that make it stronger or weaker. A HR deparment of a company might want to introduce policies that weaken it's negative effect and use possible positive effects.

At Google they got people to cancel training courses instead of no-showing by reminded people of the group image of Google and being googly. They positively used the stereotype of Googlers.