`array_api_compat.numpy.asarray(torch.Tensor)` returns a `torch.Tensor`

This feels like a bug to me. My expectation is that `xp.asarray` converts to an array object in the `xp` namespace if that operation is supported.

```
>>> import torch
>>> import array_api_compat.numpy
>>> x = torch.arange(10)
>>> array_api_compat.numpy.asarray(x)
tensor([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
```