Change Groovydoc processing to support antlr4 grammar including getting comments from AST

The new Parrot parser (currently on the master branch) supports capturing comments within the AST. Groovydoc should be adapted to utilise that capability. Currently, in SimpleGroovyClassDocAssembler#getJavaDocCommentsBeforeNode(), a crude regexp is used to extract the comment out from the source based on the relative position of a known AST node. Instead, with the new parser, we can simply extract it from the AST.