fix row identification problem in isohunt

because of their bad website which doesn't produce the same page each time

weboob/backends/isohunt/pages/torrents.py

@@ -28,7 +28,9 @@ class TorrentsPage(BasePage):
     def iter_torrents(self):
         for tr in self.document.getiterator('tr'):
             if tr.attrib.get('class', '') == 'hlRow':
-                # TODO à corriger
+                # sometimes the first tr also has the attribute hlRow
+                # i use that to ditinct it from the others
+                if tr.attrib.has_key('onmouseout'):
                     atitle = tr.getchildren()[2].getchildren()[1]
                     title = atitle.text
                     if not title: