Mediawiki reader: interpret HTML in verbatim contexts as text.

Thanks to Jeff Runningen for reporting that mediawiki works
this way.

src/Text/Pandoc/Readers/MediaWiki.hs

@@ -139,8 +139,7 @@ charsInTags tag = try $ do
   (_,raw) <- htmlTag (~== TagOpen tag [])
   if '/' `elem` raw   -- self-closing tag
      then return ""
-     else innerText . parseTags <$>
-            manyTill anyChar (htmlTag (~== TagClose tag))
+     else manyTill anyChar (htmlTag (~== TagClose tag))
 
 --
 -- main parser

tests/mediawiki-reader.wiki

@@ -105,7 +105,7 @@ H<sub>2</sub>O base<sup>''exponent''</sup>
 
 == inline code ==
 
-<code>*&rarr;*</code> <tt>typed</tt> <hask>&gt;&gt;=</hask>
+<code>*→*</code> <tt>typed</tt> <hask>>>=</hask>
 
 == code blocks ==