For PMD parsing, we use Rhino. The latest version requires at least Java 8.
Fixing this issue means, we'll need to move the javascript module to the java8-requiring modules (like apex, visualforce, scala).

See also #201 : Note, the grammar is only used for CPD tokenization. We should drop the grammar in favor of Rhino parsing, as long as we can keep the behavior for comment-based suppressions. (see #1039)