error message for `Tensor.index_put_` could be improved for MPS failure mode

[alisterburt]

import torch

device = torch.device("mps")  # Requires Apple Silicon Mac
image = torch.zeros(10, dtype=torch.complex64, device=device)
data = torch.ones(3, dtype=torch.complex64, device=device)
indices = torch.tensor([1, 3, 5], device=device)

# This fails on MPS with complex64
image.index_put_(indices=(indices,), values=data, accumulate=True)

RuntimeError: Expected supportedFloatingType(scalar_type) || scalar_type == kInt || scalar_type == kBool to be true, but got false.  (Could this error message be improved?  If so, please report an enhancement request to PyTorch.)

Ideally this error would explicitly say that index_put_ does not support complex value insertion with the MPS backend

cc @ezyang @anjali411 @dylanbespalko @mruberry @nikitaved @amjames @kulinseth @albanD @malfet @DenisVieriu97 @jhavukainen

[malfet]

True, but also no good reason why shouldn't it just support index_put for complex numbers...