Chore: remove grad from nlist linear model (#4380)

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

- **New Features**
	- Improved gradient tracking capabilities during model forward passes.
- Enhanced weight computation for atomic models, allowing for better
handling of neighbor list sizes and transitions.

- **Bug Fixes**
	- Resolved issues with weight calculations based on atom distances.
- Added error handling to ensure proper boundaries in weight
computations.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

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

deepmd/pt/model/atomic_model/linear_atomic_model.py

@@ -253,11 +253,12 @@ def forward_atomic(
         extended_coord = extended_coord.view(nframes, -1, 3)
         sorted_rcuts, sorted_sels = self._sort_rcuts_sels()
         nlists = build_multiple_neighbor_list(
-            extended_coord,
+            extended_coord.detach(),
             nlist,
             sorted_rcuts,
             sorted_sels,
         )
+
         raw_nlists = [
             nlists[get_multiple_nlist_key(rcut, sel)]
             for rcut, sel in zip(self.get_model_rcuts(), self.get_model_nsels())