Add some breathing room to file name length limit

Because Amazon stores metadata keys prefixed with `x-amz-` which might
get counted as part of the size.

app/main/views/send.py

@@ -558,7 +558,7 @@ def _check_messages(service_id, template_id, upload_id, preview_row, letters_as_
             valid=True,
             original_file_name=unicode_truncate(
                 request.args.get('original_file_name', ''),
-                1872
+                1600,
             ),
         )
     else:

tests/app/main/views/test_send.py

@@ -700,7 +700,7 @@ def test_file_name_truncated_to_fit_in_s3_metadata(
     assert sys.getsizeof(''.join((
         '{}{}'.format(key, value) for key, value in
         mock_s3_set_metadata.call_args_list[0][1].items()
-    )).encode('utf-8')) == 1998
+    )).encode('utf-8')) == 1724
 
 
 def test_show_all_columns_if_there_are_duplicate_recipient_columns(