Implement name inference rule for empty_like, clone

[zou3519]

Stack from ghstack:

• Name inference for softmax, log_softmax and Dimname overloads. #24087 Name inference for softmax, log_softmax and Dimname overloads.
• Fix out= function semantics for named tensors. #24028 Fix out= function semantics for named tensors.
• Update tensor.view_names / tensor.names_ API #23973 Update tensor.view_names / tensor.names_ API
• Rename set_names -> view_names, set_names_ -> names_ #23962 Rename set_names -> view_names, set_names_ -> names_
• Implement tensor.align_to(names), torch.align_tensors(t1, t2, ...) #23804 Implement tensor.align_to(names), torch.align_tensors(t1, t2, ...)
• Named inference for contiguous(), bernoulli variants, and dropout. #23808 Named inference for contiguous(), bernoulli variants, and dropout.
• Implement name inference rule for empty_like, clone #23746 Implement name inference rule for empty_like, clone
• Add names argument to ones, rand, randn, zeros, full; fix empty #23743 Add names argument to ones, rand, randn, zeros, full; fix empty
• Improve test_namedtensor.py with named tensor equality check #23801 Improve test_namedtensor.py with named tensor equality check
• Add name propagation for at::alias, add tensor.set_names #23624 Add name propagation for at::alias, add tensor.set_names

torch.empty_like(tensor) and tensor.clone() both propagate names to
the output tensor.

As a part of this change, I fixed the empty(..., names=) overload to
include the memory_format argument in the normal empty declaration
in native_functions.yaml.

Test Plan:

• [namedtensor ci]

gh-metadata: pytorch pytorch 23746 gh/zou3519/89/head

Differential Revision: D16647821

[facebook-github-bot]

@zou3519 merged this pull request in 16b6466.