[FIX] core: flush non-attachment binary fields when necessary

Non-attachment binary fields need to be flushed before reading their
size, since the latter relies on the database's binary size function.

Part-of: odoo/odoo#160708

odoo/addons/test_new_api/tests/test_new_fields.py

@@ -2707,6 +2707,9 @@ class TestFields(TransactionCaseWithUserDemo):
         self.assertEqual(record_bin_size.image, b'31.54 Kb')
         self.assertEqual(record_bin_size.image_512, b'1.02 Kb')
         self.assertEqual(record_bin_size.image_256, b'424.00 bytes')
+        # non-attachment binary fields: value returned as str in a different
+        # form, because coming from PostgreSQL instead of filestore
+        self.assertEqual(record_bin_size.image_64, '148 bytes')
 
         # ensure image_data_uri works (value must be bytes and not string)
         self.assertEqual(record.image_256[:8], b'iVBORw0K')

odoo/models.py

@@ -3893,11 +3893,16 @@ class BaseModel(metaclass=MetaModel):
             (column_fields if field.column_type else other_fields).add(field)
 
         # necessary to retrieve the en_US value of fields without a translation
-        translated_field_names = [field.name for field in column_fields if field.translate]
-        if translated_field_names:
-            self.flush_model(translated_field_names)
-
         context = self.env.context
+        field_names_to_flush = [
+            field.name for field in column_fields
+            if field.translate or (
+                field.type == 'binary'
+                and (context.get('bin_size') or context.get('bin_size_' + field.name))
+            )
+        ]
+        if field_names_to_flush:
+            self.flush_model(field_names_to_flush)
 
         if column_fields:
             # the query may involve several tables: we need fully-qualified names