tokenise comments

Currently comments are treated as whiteSpace. This leaves {-# LANGUAGE #-} and other Haskell pragmas out of parsed AST.

This proposal is to parse comments as a distinct token

`whiteSpace` could return [String] where each string is a separate comment line

Start by tracing the whiteSpace parser and how comments are currently discarded before the parsed AST is built. Check how {-# LANGUAGE #-} pragmas are handled, then define the token representation needed for distinct comment lines; done means comments and Haskell pragmas are preserved in the AST without being treated as whitespace.

Written by the indexing model from the issue text.