[Fusion] [Graph] Add qknorm rope fusion operator (#4711)

### What this PR does / why we need it?
This PR add `qkv_rmsnorm_rope` operator and introduces a graph fusion
pass for `qknorm_rope` operations. The implementation includes a new
configuration flag, a pattern matching pass using
`torch._inductor.pattern_matcher`, and a custom Triton kernel for the
fused operation.

Co-authored-by: Angazenn
[supperccell@163.com](mailto:supperccell@163.com)

### Does this PR introduce _any_ user-facing change?
Yes, add new additional_config

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

---------

Signed-off-by: wxsIcey <1790571317@qq.com>

vllm_ascend/spec_decode/eagle_proposer.py

@@ -25,6 +25,7 @@ from vllm_ascend.ascend_forward_context import set_ascend_forward_context
 from vllm_ascend.attention.attention_mask import AttentionMaskBuilder
 from vllm_ascend.attention.attention_v1 import AscendAttentionState
 from vllm_ascend.attention.utils import AscendCommonAttentionMetadata
+from vllm_ascend.ops.rotary_embedding import update_cos_sin
 from vllm_ascend.spec_decode.interface import Proposer, SpecDcodeType
 
 PADDING_SLOT_ID = -1
@@ -143,6 +144,9 @@ class EagleProposer(Proposer):
                   aclgraph_runtime_mode: CUDAGraphMode = CUDAGraphMode.NONE,
                   batch_descriptor=None,
                   dummy_compute_logits=lambda hidden_states: None):
+        # update global cos, sin
+        update_cos_sin(self.positions[:num_tokens])
+
         with set_ascend_forward_context(None,
                                         self.vllm_config,
                                         num_tokens=num_tokens):
@@ -338,6 +342,8 @@ class EagleProposer(Proposer):
         builder = self.runner.attn_groups[0][0].get_metadata_builder()
         attn_metadata = builder.build(0, common_attn_metadata,
                                       self.runner.get_model())
+        # update global cos, sin
+        update_cos_sin(self.positions[:num_input_tokens])
 
         with set_ascend_forward_context(attn_metadata,
                                         self.vllm_config,
@@ -443,6 +449,10 @@ class EagleProposer(Proposer):
 
             attn_metadata.attn_mask = attn_mask
             # Run the model.
+
+            # update global cos, sin
+            update_cos_sin(self.positions[:input_batch_size])
+
             with set_ascend_forward_context(attn_metadata,
                                             self.vllm_config,
                                             num_tokens=input_batch_size):