Add failing test when file contains a large amount of comments

[jfmengels]

This creates a stack overflow. I am not sure how to fix it, but I have created a regression test, should anyone want to look into this 🙏

[lue-bird]

for reference, I tried replacing our whitespace/comments/empty primitive with a loop and it got ~10ms slower.
What I think we could do though is only start looping once we encounter a single-line comment (might be a bit rough though)

[jfmengels]

I think that would make sense, although I'm not sure what you mean by rough.