I suspect this might also fix a number of other bugs around rules too-greedily warning on non-components. Would you mind checking the open issues and seeing if there's additional test cases that could apply here?

Oh yes, I remember at least seeing #1800 which seems to be the same issue that's being fixed here. Will check that and others.

Added tests for the original issue in #1800, plus #1800 (comment)
I had high hopes for #512, but since that returns createElement, it gets a non-zero confidence value.
(and Travis failure is a fluke again)