chore(pt): Change the type of do_message_passing from int to bool in DeepPotPT and DeepSpinPT classes (#4391)

Fix #4366.

* Update the type of `do_message_passing` to `bool` in the `DeepPotPT`
class and `init` method in `source/api_cc/include/DeepPotPT.h` and
`source/api_cc/src/DeepPotPT.cc`
* Update the type of `do_message_passing` to `bool` in the `DeepSpinPT`
class and `init` method in `source/api_cc/include/DeepSpinPT.h` and
`source/api_cc/src/DeepSpinPT.cc`

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

## Summary by CodeRabbit

- **New Features**
- Enhanced error handling for exceptions from the PyTorch library in
both `DeepPotPT` and `DeepSpinPT` classes.
- Simplified boolean checks for message passing in the `compute` methods
of both classes.

- **Bug Fixes**
- Improved robustness in the `DeepPotPT` constructor to prevent resource
leaks during initialization.

- **Documentation**
- Updated method signatures to reflect changes in parameter types and
structures.

<!-- 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>

Author: njzjz

source/api_cc/include/DeepPotPT.h

@@ -335,7 +335,7 @@ class DeepPotPT : public DeepPotBackend {
   NeighborListData nlist_data;
   int max_num_neighbors;
   int gpu_id;
-  int do_message_passing;  // 1:dpa2 model 0:others
+  bool do_message_passing;  // 1:dpa2 model 0:others
   bool gpu_enabled;
   at::Tensor firstneigh_tensor;
   c10::optional<torch::Tensor> mapping_tensor;

source/api_cc/include/DeepSpinPT.h

@@ -257,7 +257,7 @@ class DeepSpinPT : public DeepSpinBackend {
   NeighborListData nlist_data;
   int max_num_neighbors;
   int gpu_id;
-  int do_message_passing;  // 1:dpa2 model 0:others
+  bool do_message_passing;  // 1:dpa2 model 0:others
   bool gpu_enabled;
   at::Tensor firstneigh_tensor;
   c10::optional<torch::Tensor> mapping_tensor;

source/api_cc/src/DeepPotPT.cc

@@ -171,7 +171,7 @@ void DeepPotPT::compute(ENERGYVTYPE& ener,
     nlist_data.copy_from_nlist(lmp_list);
     nlist_data.shuffle_exclude_empty(fwd_map);
     nlist_data.padding();
-    if (do_message_passing == 1) {
+    if (do_message_passing) {
       int nswap = lmp_list.nswap;
       torch::Tensor sendproc_tensor =
           torch::from_blob(lmp_list.sendproc, {nswap}, int32_option);
@@ -234,7 +234,7 @@ void DeepPotPT::compute(ENERGYVTYPE& ener,
             .to(device);
   }
   c10::Dict<c10::IValue, c10::IValue> outputs =
-      (do_message_passing == 1)
+      (do_message_passing)
           ? module
                 .run_method("forward_lower", coord_wrapped_Tensor, atype_Tensor,
                             firstneigh_tensor, mapping_tensor, fparam_tensor,

source/api_cc/src/DeepSpinPT.cc

@@ -179,7 +179,7 @@ void DeepSpinPT::compute(ENERGYVTYPE& ener,
     nlist_data.copy_from_nlist(lmp_list);
     nlist_data.shuffle_exclude_empty(fwd_map);
     nlist_data.padding();
-    if (do_message_passing == 1) {
+    if (do_message_passing) {
       int nswap = lmp_list.nswap;
       torch::Tensor sendproc_tensor =
           torch::from_blob(lmp_list.sendproc, {nswap}, int32_option);
@@ -234,7 +234,7 @@ void DeepSpinPT::compute(ENERGYVTYPE& ener,
             .to(device);
   }
   c10::Dict<c10::IValue, c10::IValue> outputs =
-      (do_message_passing == 1)
+      (do_message_passing)
           ? module
                 .run_method("forward_lower", coord_wrapped_Tensor, atype_Tensor,
                             spin_wrapped_Tensor, firstneigh_tensor,