Yeah that was my very first thought but it's not the case. In fact in the latest doc where it occurs, there are 3 properly closed pairs of parens on the same page, just south of where the macro surrenders & heads home.

(A separate routine runs on each file beforehand which, among other things, ensures all parens are paired.)