Description:
------------
When converting a hexadecimal html entity to UTF-8 with mb_convert_encoding, it get's converted to a broken unicode character (displayed in firefox as a small square). 

Reproduce code:
---------------
$v_html = "š";
echo $v_html;
echo mb_convert_encoding($v_html, 'UTF-8', 'HTML-ENTITIES');
echo html_entity_decode($v_html, ENT_COMPAT, 'UTF-8');

Expected result:
----------------
I'd expect it to output three times the same character, š.

At first as hexadecimal html entity, then two times in UTF-8.

š??

Actual result:
--------------
š[something broken]?