[FreshRSS] decode special chars for MAYBE_HTML

Patch for simplepie#350

It is quite common for feeds (e.g. Facebook as reported above) to have
some MAYBE_HTML section HTML-encoded, which is currently not handled by
SimplePie
```xml
<title><![CDATA[ L&simplepie#39;alpha 11 est arriv&#xe9;e...]]></title>
```

This is the approach currently used (with success) in FreshRSS:
FreshRSS/FreshRSS#754
FreshRSS/FreshRSS#813

library/SimplePie/Sanitize.php

@@ -229,6 +229,7 @@ public function sanitize($data, $type, $base = '')
 		{
 			if ($type & SIMPLEPIE_CONSTRUCT_MAYBE_HTML)
 			{
+				$data = htmlspecialchars_decode($data, ENT_QUOTES); 
 				if (preg_match('/(&(#(x[0-9a-fA-F]+|[0-9]+)|[a-zA-Z0-9]+)|<\/[A-Za-z][^\x09\x0A\x0B\x0C\x0D\x20\x2F\x3E]*' . SIMPLEPIE_PCRE_HTML_ATTRIBUTE . '>)/', $data))
 				{
 					$type |= SIMPLEPIE_CONSTRUCT_HTML;