add xpu device index for Int8Params

[faaany]

The following code will not move tensor to the desired target_device:

import torch
import bitsandbytes as bnb

random_tensor = torch.randn((3, 3), dtype=torch.float16)
target_device = 2
new_value = bnb.nn.Int8Params(random_tensor, requires_grad=False).to(target_device)
print(new_value)
# Parameter containing:
# Parameter(Int8Params([[ 127,  -97, -126],
#             [  60,   11,  127],
#             [-127,   90,  -35]], device='xpu:0', dtype=torch.int8))

To fix this, we need to pass the device index to the xpu method just like in the cuda method.

[yao-matrix]

@jiqing-feng

[github-actions]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[matthewdouglas]

LGTM, thanks!