Why 100% test coverage proves nothing

This is the follow-up to my previous post.

I wrote an example - obviously badly written on purpose - by which I prove why 100% code coverage is useless.

While these tests cover 100%, they miss a test case that I added in the comments on the bottom.