[IMP] models: faster browse records iteration

By avoiding to update the prefetch ids for each record when iterating on
recordsets.  This is faster when iterating on very large recordsets.

odoo/models.py

@@ -4314,7 +4314,7 @@ class BaseModel(MetaModel('DummyModel', (object,), {'_register': False})):
     #
 
     @classmethod
-    def _browse(cls, ids, env, prefetch=None):
+    def _browse(cls, ids, env, prefetch=None, add_prefetch=True):
         """ Create a recordset instance.
 
         :param ids: a tuple of record ids
@@ -4327,7 +4327,8 @@ class BaseModel(MetaModel('DummyModel', (object,), {'_register': False})):
         if prefetch is None:
             prefetch = defaultdict(set)         # {model_name: set(ids)}
         records._prefetch = prefetch
-        prefetch[cls._name].update(ids)
+        if add_prefetch:
+            prefetch[cls._name].update(ids)
         return records
 
     def browse(self, arg=None, prefetch=None):
@@ -4634,7 +4635,7 @@ class BaseModel(MetaModel('DummyModel', (object,), {'_register': False})):
     def __iter__(self):
         """ Return an iterator over ``self``. """
         for id in self._ids:
-            yield self._browse((id,), self.env, self._prefetch)
+            yield self._browse((id,), self.env, self._prefetch, False)
 
     def __contains__(self, item):
         """ Test whether ``item`` (record or field name) is an element of ``self``.