[Misc] Add ACL_OP_INIT_MODE env var and set default to 1 (#597)

### What this PR does / why we need it?
Fix the bug in torch 2.5.1 that raising segment fault when enable
`pin_memory` while creating a tensor using `torch.tensor`.

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

---------

Signed-off-by: shen-shanshan <467638484@qq.com>

vllm_ascend/envs.py

@@ -112,6 +112,15 @@ env_variables: Dict[str, Callable[[], Any]] = {
     "VLLM_ASCEND_MODEL_EXECUTE_TIME_OBSERVE":
     lambda: bool(int(os.getenv("VLLM_ASCEND_MODEL_EXECUTE_TIME_OBSERVE", '0'))
                  ),
+    # VLLM_ASCEND_ACL_OP_INIT_MODE:
+    #   0: default, normal init.
+    #   1: delay init until launch aclops.
+    #   2: forbid aclops init and launch.
+    # Find more details at https://gitee.com/ascend/pytorch/pulls/18094
+    # We set this var default to `1` in vllm-ascend to avoid segment fault when
+    # enable `pin_memory` while creating a tensor using `torch.tensor`.
+    "VLLM_ASCEND_ACL_OP_INIT_MODE":
+    lambda: os.getenv("VLLM_ASCEND_ACL_OP_INIT_MODE", '1'),
 }
 
 # end-env-vars-definition

vllm_ascend/platform.py

@@ -24,6 +24,7 @@ import vllm.envs as envs
 from vllm.logger import logger
 from vllm.platforms import Platform, PlatformEnum
 
+import vllm_ascend.envs as ascend_envs
 from vllm_ascend.ascend_config import check_ascend_config, init_ascend_config
 from vllm_ascend.utils import ASCEND_QUATIZATION_METHOD, update_aclgraph_sizes
 
@@ -46,6 +47,7 @@ else:
     FlexibleArgumentParser = None
 
 os.environ["RAY_EXPERIMENTAL_NOSET_ASCEND_RT_VISIBLE_DEVICES"] = "1"
+os.environ["ACL_OP_INIT_MODE"] = ascend_envs.VLLM_ASCEND_ACL_OP_INIT_MODE
 
 
 class NPUPlatform(Platform):