I fail to see how paradigm is not orthogonal to testing. Come to think about, I wrote my first unit test in Erlang. That’s after years of C++, C#, Python and such. What was crucial for tests v.s. no tests, was the industry. We didn’t cover much in game dev, because in the game engine pieces of code should be either trivial or malleable. Either way tests didn’t contribute much. It was simply cheaper not to test as the benefit would be minuscule. But in back-end it is best to have your code covered because you want predictability over malleability. It’s cheaper to write test given such priorities.
Functional code is indeed easier to cover with unit-tests, but shouldn’t this be a pro argument?