[BugFix] Fix multimodal model support fullgraph error (#3425)

### What this PR does / why we need it?
Because the update_attn_params function requires passing the num_tokens
parameter, and num_tokens is obtained via postions.shape[0]. However,
the multimodal model uses mrope (Multidimensional Rotary Position
Embedding), which results in the postions having a shape of 2.
Consequently, postions.shape[0] retrieves an incorrect value.We resolve
this issue by replacing positions.shape[0] with maybe_padded_num_tokens.

- vLLM version: v0.11.0rc3
- vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0

Signed-off-by: fan2956 <zhoufan53@huawei.com>

vllm_ascend/worker/model_runner_v1.py

@@ -1520,13 +1520,14 @@ class NPUModelRunner(LoRAModelRunnerMixin):
 
         forward_context = get_forward_context()
         if forward_context.cudagraph_runtime_mode == CUDAGraphMode.FULL:
+            # TODO: maybe_padded_num_tokens will be removed, use num_input_tokens instead
             if self.vllm_config.model_config.use_mla:
                 # FIXME: Try using `auto_dispatch_capture=True`
                 update_mla_attn_params(self.update_stream, forward_context,
-                                       positions.shape[0])
+                                       maybe_padded_num_tokens)
             else:
                 update_attn_params(self.update_stream, forward_context,
-                                   positions.shape[0])
+                                   maybe_padded_num_tokens)
 
         if get_forward_context().sp_enabled:
             hidden_states = tensor_model_parallel_all_gather(hidden_states, 0)