Kata

Your random tests are vulnerable to input modification ( I could not submit, but I think you can still see my solution by clicking under this comment ), and your submit test suite is missing fixed tests ( should have had the example tests as a section of the submit tests ).

This would [ very probably ] have been pointed out if you had posted a link to Discord#reviewing before publishing.

In general, the idea was apparently ( somewhat ) better than the previous one, but still not good enough. Inviting the use of eval in solutions is often frowned upon as well.

If people don't subtitle their vote, it's impossible to know for sure what killed the kata, but you're still dropping the ball on a number of things people could have told you in advance.