[AQT-bug] AffineQuantizedTensor raises error with .cuda()

[gau-nernst]

import torch
from torchao.quantization.quant_api import quantize_, int8_weight_only

model = torch.nn.Linear(2048, 2048)
quantize_(model, int8_weight_only())
# model.to("cuda")  # this works
model.cuda()  # this doesn't

  File "/home/xx/code/ao/torchao/dtypes/affine_quantized_tensor.py", line 975, in _
    args[0].to(*args[1:], **kwargs)._apply_fn_to_data(torch.clone),
  File "/home/xx/code/ao/torchao/dtypes/affine_quantized_tensor.py", line 291, in to
    kwargs = self._get_to_kwargs(*args, **kwargs)
  File "/home/xx/code/ao/torchao/dtypes/affine_quantized_tensor.py", line 277, in _get_to_kwargs
    device, dtype, _, memory_format = torch._C._nn._parse_to(*args, **kwargs)
TypeError: to() received an invalid combination of arguments - got (device=torch.device, layout=torch.layout, dtype=torch.dtype, ), but expected one of:
 * (torch.device device = None, torch.dtype dtype = None, bool non_blocking = False, bool copy = False, *, torch.memory_format memory_format = None)
 * (torch.dtype dtype, bool non_blocking = False, bool copy = False, *, torch.memory_format memory_format = None)
 * (Tensor tensor, bool non_blocking = False, bool copy = False, *, torch.memory_format memory_format = None)

I faced similar problems with other subclasses that I implemented before. Basically need to remove the layout kwarg.

cc: @jerryzh168

[jerryzh168]

thanks, yeah I think we want to add some basic tests to https://github.com/pytorch/ao/blob/main/test/dtypes/test_affine_quantized.py