Support negation/subtraction of nested_tensor

[frankier]

🚀 The feature, motivation and pitch

These operations are not currently supported. Since add is supported, the implementation might not be too much work.

>>> - torch.nested.nested_tensor([torch.tensor([-2.3, 0.4, 5.3]), torch.tensor([-2.3, 0.4, 5.3])])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NotImplementedError: Could not run 'aten::neg' with arguments from the 'NestedTensorCPU' backend. This could be because the operator doesn't exist for this backend, or was omitted during the selective/custom build process (if using custom build). If you are a Facebook employee using PyTorch on mobile, please visit https://fburl.com/ptmfixes for possible resolutions. 'aten::neg' is only available for these backends: [CPU, CUDA, HIP, MPS, IPU, XPU, HPU, VE, Meta, PrivateUse1, PrivateUse2, PrivateUse3, FPGA, ORT, Vulkan, Metal, QuantizedCPU, QuantizedCUDA, QuantizedHIP, QuantizedMPS, QuantizedIPU, QuantizedXPU, QuantizedHPU, QuantizedVE, QuantizedMeta, QuantizedPrivateUse1, QuantizedPrivateUse2, QuantizedPrivateUse3, CustomRNGKeyId, MkldnnCPU, SparseCPU, SparseCsrCPU, SparseCsrCUDA, BackendSelect, Python, FuncTorchDynamicLayerBackMode, Functionalize, Named, Conjugate, Negative, ZeroTensor, ADInplaceOrView, AutogradOther, AutogradCPU, AutogradCUDA, AutogradHIP, AutogradXLA, AutogradMPS, AutogradIPU, AutogradXPU, AutogradHPU, AutogradVE, AutogradLazy, AutogradMeta, AutogradPrivateUse1, AutogradPrivateUse2, AutogradPrivateUse3, AutogradNestedTensor, Tracer, AutocastCPU, AutocastCUDA, FuncTorchBatched, FuncTorchVmapMode, Batched, VmapMode, FuncTorchGradWrapper, PythonTLSSnapshot, FuncTorchDynamicLayerFrontMode, PythonDispatcher].

Alternatives

Depending on the situation, it may be possible to negate all values before putting them into the nested_tensor. Or else unbind() could be used and work inner tensor at a time.

Additional context

No response

cc @cpuhrsch @jbschlosser @bhosmer @drisspg @mikaylagawarecki

[cpuhrsch]

@frankier - we just landed support for neg. It should become available in the nightlies within the next 48 hours. Can you just confirm that this is what you wanted and otherwise please reopen the issue? Thanks