drop ascend scheduler (#4498)

Ascend scheduler was added for non chunk prefill case before, since that
the npu ops didn't work well with chunked prefill.

Now the ops with chunked prefill work better, it's time to remove the
ascend scheduler to use vLLM default scheduler.

- vLLM version: v0.11.2

---------

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

vllm_ascend/worker/model_runner_v1.py

@@ -330,10 +330,6 @@ class NPUModelRunner(LoRAModelRunnerMixin):
 
         # Ascend-specific configurations
         self.ascend_config = get_ascend_config()
-        if self.ascend_config.ascend_scheduler_config.enabled:
-            self.chunked_prefill_enabled = self.scheduler_config.chunked_prefill_enabled
-        else:
-            self.chunked_prefill_enabled = True
         self.weight_prefetch_method = WeightPrefetchMethod(
             self.ascend_config.weight_prefetch_config)
         # Dump / PrecisionDebugger configuration now comes from AscendConfig
@@ -1942,7 +1938,6 @@ class NPUModelRunner(LoRAModelRunnerMixin):
 
     def _build_attn_state(self, num_reqs, num_scheduled_tokens,
                           num_valid_tokens):
-        ascend_config = get_ascend_config()
         if np.array_equal(self.seq_lens_np[:num_reqs], num_scheduled_tokens):
             attn_state = AscendAttentionState.PrefillNoCache
         # We assume it is the decode stage, where prefill occurs but only one token is not hit in cache.
@@ -1959,7 +1954,7 @@ class NPUModelRunner(LoRAModelRunnerMixin):
             else:
                 attn_state = AscendAttentionState.ChunkedPrefill
         # splitfuse
-        elif not ascend_config.ascend_scheduler_config.enabled or self.chunked_prefill_enabled:
+        elif self.scheduler_config.enable_chunked_prefill:
             attn_state = AscendAttentionState.ChunkedPrefill
         else:
             attn_state = AscendAttentionState.PrefillCacheHit