The Haskell version isn't in-place and therefore not really quicksort, agreed, but that's a separate (though valid) criticism. It doesn't make a "straw man" of the Java version, does it? It would be a straw man if it said "here, look at a reasonable quicksort implementation in Java (absurd, bloated code follows)".
The Java version doesn't really have a lot of unnecessary fluff. What, it's not a static method and has instance variables? So what? That doesn't add a lot of verbosity and is NOT the crux of the author's argument either.
The author was doing a section by section comparison of the two - Look! There's no Haskell needed for the corresponding Java code! He is deliberately showing verbosity in Java with an apple-to-orange strawman comparison. What else is he trying to show?
The instance variable in class is an important strawman the author added to Java. He's trying to show the need of "state" in Java, which is not needed in a sensible Java version of qsort, as all data can be passed in parameters.
He also made the statement that the instance variable is needed for recursion in Java (!) to "substantiate" (make up) the excuse for using instance variable in Java.
The Java version doesn't really have a lot of unnecessary fluff. What, it's not a static method and has instance variables? So what? That doesn't add a lot of verbosity and is NOT the crux of the author's argument either.