handle code conflict

Signed-off-by: weijinqian_v1 <weijinqian@huawei.com>

Author: weijinqian_v1

vllm_ascend/ascend_forward_context.py

@@ -103,7 +103,7 @@ def set_ascend_forward_context(
                 max_tokens_across_dp / tp_world_size) * world_size
 
             if fused_moe_state == FusedMoEState.MC2_PREFILL:
-                chunk_size = envs.VLLM_ASCEND_FUSED_MOE_MC2_CHUNK_SIZE
+                chunk_size = envs_ascend.VLLM_ASCEND_FUSED_MOE_MC2_CHUNK_SIZE
                 forward_context.max_num_chunks = math.ceil(
                     math.ceil(max_tokens_across_dp / tp_world_size) /
                     chunk_size)