Prior to this change, `XpathResultMatchers` and more generally the
`MockHttpServletResponse` would default to ISO-8859-1 encoding even when
it's not supposed to. The Servlet/HTTP specs mention this encoding
for all `text/*` mime types when decoding bodies to Strings, but this
issue is about XML Parsers.
XML Parsers should use the encoding:
* defined in the `Content-Type` response header (if available)
* written in the XML declaration of the document
* "guessed" by a built-in auto-detection mechanism
This commit changes the following:
* XPathMatchers now feed the XML parser with byte arrays instead of
decoded Strings
* the response should be written to `MockHttpServletResponse` using
its OutputStream, and not a PrintWriter which defaults to ISO-8859-1
Issue: SPR-12676