fix the unittest bug of none grad of margin_cross_entropy when FLAGS_retain_grad_for_all_tensor change default setting (#43241)

* use yapf to format all python file

* yapf exclude two unittests file for they rely on writing and reading file, and format will break them

* disable diff_py_file because too many diff files cause command following failed

* support margin loss (arcface, cosface, sphereface)