DISCLAIMER - Failure to meet any of these points doesn't mean new ideas or pull requests should be outright rejected. They should be used as points of discussion to ultimately help us ship code faster.
code is used in reference to the code being tested. The target, maybe?
test is used in reference to the code that is exercising the codebase to validate correctness.
If a project has a de-facto standard, it should be maintained. The process for changing a de-facto standard should be people focused, and should not be initially communicated via a pull request.
If the test behaviour isn't obvious by looking at it, it's no good. If I have to scroll for days to read a test, it's no good.
If I make a single code change and multiple tests fail, it's no good.
If I can delete code that is needed and it's corresponding test suite still passes, it's no good.
If I break a feature and a test fails, but it isn't obvious what is wrong, it's no good.
If a test inconsistently fails, it's no good.
If a single test takes more than 2 seconds to run, it's no good. Exceptions can be allowed for Selenium.
If a test can't be described by the impact it has on the consumer of the code, it probably doesn't warrant a test.