Description:
------------
mb_convert_encoding converts symbols \xAD\xB5-\xAD\xBF  incorrectly 
from EUC-JP to UTF-8. It's possible that some other symbols converted 
incorrectly too, but I have no possibility to check it to full 
extent.

Unicode has corresponding codepoints, i.e. U+2161 for Ⅱ.

Majority of EUC-JP texts is converted mormally.

Reproduce code:
---------------
echo mb_convert_encoding("\xAD\xB6", "UTF-8", "EUC-JP");

Expected result:
----------------
string ?Ⅱ? (U+2161)
printed to STDOUT

Actual result:
--------------
string ???
printed to STDOUT