[lang] Fix MPS / other torch backend from_torch (#8298)

Issue: #6861

### Brief Summary

<!--
copilot:summary
-->
### <samp>🤖 Generated by Copilot at fd38fea</samp>

Fix a bug that prevents using torch tensors on non-CPU devices other
than CUDA in Taichi kernels. Update the condition in `kernel_impl.py` to
copy tensors to the correct device before and after kernel execution.

### Walkthrough

<!--
copilot:walkthrough
-->
### <samp>🤖 Generated by Copilot at fd38fea</samp>

* Fix a bug that prevents using torch tensors on non-CPU devices other
than CUDA
([link](https://github.com/taichi-dev/taichi/pull/8298/files?diff=unified&w=0#diff-a157043b38542c8145447ff342fda65fe4d54fb777fe514daa70007e83e20dc1L752-R752))

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

python/taichi/lang/kernel_impl.py

@@ -749,7 +749,9 @@ def call_back():
                         v.grad = torch.zeros_like(v)
 
                     tmp = v
-                    if str(v.device).startswith("cuda") and taichi_arch != _ti_core.Arch.cuda:
+                    if (str(v.device) != "cpu") and not (
+                        str(v.device).startswith("cuda") and taichi_arch == _ti_core.Arch.cuda
+                    ):
                         # Getting a torch CUDA tensor on Taichi non-cuda arch:
                         # We just replace it with a CPU tensor and by the end of kernel execution we'll use the
                         # callback to copy the values back to the original CUDA tensor.