rhino xhr does not use contentType;charset or other content sniffing

In env-js/src/platform/rhino @ line 251

    xhr.responseText = java.nio.charset.Charset.forName("UTF-8").
        decode(java.nio.ByteBuffer.wrap(baos.toByteArray())).toString()+"";

will either explode with an exception or produce junk if the doco is in latin-1 (or equiv)

First step is to write a test to prove this. then ideally follow a basic charset sniffer algorithm:

• use content-type;charset=
• if html, look for meta tag
• look at byte-order marks
• default to utf8

there are more complicated versions to do this, but this should cover the basics

In a different note, a bit above (line 219)

   var contentEncoding = connection.getContentEncoding() || "utf-8",

should just be || '', the content-encoding isn't used for charset encoding.