fix(paddle): Revert strict=True to strict=False in zip() calls

- Reverted all zip() strict=True back to strict=False due to Paddle JIT incompatibility
- Affected files: deep_eval.py, se_a.py, transform_output.py, dataloader.py, nlist.py
- Fixes TypeError: convert_zip() got an unexpected keyword argument 'strict'

The strict parameter is not supported in Paddle's JIT-compiled code, causing test failures in test_dpa2.py and potentially other tests. Reverting to strict=False resolves this compatibility issue.

Co-authored-by: njzjz <9496702+njzjz@users.noreply.github.com>

Author: Copilot

deepmd/pd/infer/deep_eval.py

@@ -385,7 +385,7 @@ def eval(
             zip(
                 [x.name for x in request_defs],
                 out,
-                strict=True,
+                strict=False,
             )
         )
 

deepmd/pd/model/descriptor/se_a.py

@@ -769,7 +769,7 @@ def forward(
                 self.filter_layers.networks,
                 self.compress_data,
                 self.compress_info,
-                strict=True,
+                strict=False,
             )
         ):
             if self.type_one_side:

deepmd/pd/model/model/transform_output.py

@@ -122,7 +122,7 @@ def take_deriv(
     split_vv1 = paddle.split(vv1, [1] * size, axis=-1)
     split_svv1 = paddle.split(svv1, [1] * size, axis=-1)
     split_ff, split_avir = [], []
-    for vvi, svvi in zip(split_vv1, split_svv1, strict=True):
+    for vvi, svvi in zip(split_vv1, split_svv1, strict=False):
         # nf x nloc x 3, nf x nloc x 9
         ffi, aviri = task_deriv_one(
             vvi,

deepmd/pd/utils/dataloader.py

@@ -171,7 +171,7 @@ def construct_dataset(system: str) -> DeepmdDataSetForLoader:
         else:
             self.batch_sizes = batch_size * np.ones(len(systems), dtype=int)
         assert len(self.systems) == len(self.batch_sizes)
-        for system, batch_size in zip(self.systems, self.batch_sizes, strict=True):
+        for system, batch_size in zip(self.systems, self.batch_sizes, strict=False):
             if dist.is_available() and dist.is_initialized():
                 system_batch_sampler = DistributedBatchSampler(
                     system,

deepmd/pd/utils/nlist.py

@@ -404,7 +404,7 @@ def build_multiple_neighbor_list(
     rr.masked_fill(nlist_mask, float("inf"))
     nlist0 = nlist
     ret = {}
-    for rc, ns in zip(rcuts[::-1], nsels[::-1], strict=True):
+    for rc, ns in zip(rcuts[::-1], nsels[::-1], strict=False):
         nlist0 = nlist0[:, :, :ns].masked_fill(rr[:, :, :ns] > rc, -1)
         ret[get_multiple_nlist_key(rc, ns)] = nlist0
     return ret