adjust unit tests for test_save_load_float16

[kaixuanliu]

When we run unit test like pytest -rA tests/pipelines/wan/test_wan_22.py::Wan22PipelineFastTests::test_save_load_float16, we found that the pipeline runs w/ all fp16 datatype, but after save and reload, some parts of text-encoder in pipe_loaded uses fp32, although we set torch_dtype to fp16 explicitly. Deep investigation found that the root cause is here: L783. Here we made an adjustment to the test case to manually add the component = component.to(torch_device).half() operation to align excatly with the behavior in pipe

[kaixuanliu]

@a-r-r-o-w @DN6 pls help review, thx!

[regisss]

Not sure I understand the issue here. This specific T5 module is kept in fp32 on purpose, why forcing a fp16 cast in the test?

[kaixuanliu]

@regisss Hi, the purpose of this test case is to compare the output of pipelines using fp16 dtype(pipe) and the output of pipelines loaded from previously saved(pipe_loaded), they should be the same. However, all components of pipe is set to fp16 dtype in L1424~L1426, while for pipe_loaded, some parts are kept in fp32, which does not match exactly with the computation in pipe fwd.