Revert earlier change and use to_numpy

Revert pandas-dev#24048 change that caused bug.

Author: samuelsinayoko

pandas/core/frame.py

@@ -4615,7 +4615,7 @@ def dropna(self, axis=0, how='any', thresh=None, subset=None,
                 else:
                     raise TypeError('must specify how or thresh')
 
-            result = self.loc(axis=axis)[mask]
+            result = self._take(mask.nonzero()[0], axis=axis)
 
         if inplace:
             self._update_inplace(result)