Improve unittest.

Before:
It was two steps
1) Acquire the method (for example __iadd__).
   On a C-Level this triggers Base_getattro
2) Then call the return object.

However this is not exactly what CPython does on a statement like this:
a += 1

CPython looks first at the number protocol of that type.
If the number protocol has the slot defined it calls it, otherwise it
raises a TypeError.
https://docs.python.org/3.6/c-api/typeobj.html?highlight=pynumbermethods#c.PyNumberMethods
=> CPython omits Base_getattro at all.

Now: By using the operator module, the unittests behave exactly as
cpython on a statement like this:
a += 1

src/Acquisition/tests.py

@@ -3382,7 +3382,7 @@ def _check_special_methods(self, base_class=Acquisition.Implicit):
         # Check that special methods are proxied
         # when called implicitly by the interpreter
 
-        def binary_acquired_func(self, other):
+        def binary_acquired_func(self, other, modulo=None):
             return self.value
 
         def unary_acquired_func(self):
@@ -3426,9 +3426,11 @@ class B(Acquisition.Implicit):
         _found_at_least_one_div = False
 
         for meth in self.__binary_numeric_methods__:
-            # called on the instance
-            self.assertEqual(base.value,
-                             getattr(base.derived, meth)(-1))
+            op = getattr(operator, meth, None)
+            if op is not None:
+                # called on the instance
+                self.assertEqual(base.value, op(base.derived, -1))
+
             # called on the type, as the interpreter does
             # Note that the C version can only implement either __truediv__
             # or __div__, not both