When I started thinking about this I realized that testing for rationality is pretty complicated! The hardest part about it is determining the 'most rational person' in a group. If the 'most rational person' is a member of the group being tested, how can the testers determine who they are if the testers are less rational than them? Does a tester's ability to recognize the best of the test group depend on whether the tester is biased, and how they are biased? And who would test the testers, then?

Regardless, here's an idea or two.

A Multilevel test: Biases may be fairly easy to test for, in general. The are relatively well defined. Someone who is known to be fairly unbiased in one respect or another could run tests for that bias.

An experimental test: Learning ability could be tested. A list of 100 skills, facts and methods could be presented to individuals in a test group. Some of the items on the list would be false, untenable or in some other way illogical. Members of the group would have to learn the useful ones, thus demonstrating their ability to overcome any biases they may have to learning this-or-that concept, trait, skill or whatever. They would also have to not learn the bogus ones, demonstrating their ability to recognize bad ideas, harmful methods and false facts, etc. Groups that did well in the experiment would could be held up as examples of how one ought to approach learning.