Scientific notation is not correctly handled for complex dtypes #38285
Description
>>> torch.tensor([2.3+4j, 1e-28j])
tensor([(2.3000+4.0000j), (0.0000+0.0000j)])
>>> torch.tensor([2.3+4j, 1e-28])
tensor([(2.3000+4.0000j), (0.0000+0.0000j)])
>>> torch.tensor([2.3, 1e-28])
tensor([2.3000e+00, 1.0000e-28])
Tests similar to the tests for floating point dtypes in test_print should be added for complex dtypes as well.
cc @ezyang @anjali411 @dylanbespalko