torch_frontend.Tensor.is_conj

[Sam-Armstrong]

torch.Tensor.is_conj() needs to be implemented in the torch frontend in order to write a frontend for torch.resolve_conj.

cc #26935

[yugborana]

Hey @Sam-Armstrong I would like to work on this issue.
This is the code I would like to propose for the above issue...

def is_conj(input):
"""
Returns a single boolean indicating whether the entire input tensor represents conjugate values.

If the input tensor has an explicit `is_conj` attribute, this function will simply return it.
Otherwise, it will try to infer the conjugacy by checking the imaginary component of complex tensors.
"""
if hasattr(input, 'is_conj'):
    # If the tensor has an explicit `is_conj` attribute, use it
    return input.is_conj
else:
    # If no explicit conjugacy state, infer it from the tensor data
    if torch.is_complex(input):
        # For complex tensors, check if the imaginary component is zero for all elements
        return torch.all(torch.abs(input.imag) < 1e-8)
    else:
        # For real tensors, all elements are conjugate
        return True

If the proposed solution is correct, should I raise the PR?