Handle PI with no value in xml parser

- an XML PI contains a target and optionally some content (see https://en.wikipedia.org/wiki/Processing_Instruction)
  - the parser expected to always have some content and so it could lead to wrong parsing.

src/core/xml_parser.js

@@ -145,6 +145,7 @@ class XMLParserBase {
       pos < s.length &&
       !isWhitespace(s, pos) &&
       s[pos] !== ">" &&
+      s[pos] !== "?" &&
       s[pos] !== "/"
     ) {
       ++pos;