[Feature] Support DSA-CP for Hybrid scenario (#5702)

Signed-off-by: zzhx1 <zzh_201018@outlook.com>

### What this PR does / why we need it?
> Extracted from PR #5513
Based on the Sharded-CP feature PR:#4702;
RFC:https://github.com/vllm-project/vllm/issues/30055

### Support FULL_DECODE_ONLY Mode under PD-Mixed Scenario:
Extends DSA-CP to handle the FULL_DECODE_ONLY execution mode when
running in a prefill-decode mixed (PD-mixed) serving environment,
improving throughput and resource utilization for decode-intensive
workloads.
**In pure prefill nodes:**
- Both q_proj and o_proj are sharded across world ranks, using
**broadcast** for weights distribution.

**In PD-mixed nodes (supporting both prefill and decode):**

- q_proj is fully replicated (not sharded) to avoid communication
overhead during decoding.
- o_proj Using the original TP `RowParallelLinear` method to store
weights

**During prefill execution:**
- o_proj forwards through all_gather to collect weights, reconstructing
the complete o_proj weights on each card.

**During decode (graph replay phase):**
- Additional all_to_all (before o_proj) and reduce_scatter (after
o_proj) are introduced to enable sequence-parallel output aggregation
while maintaining correctness under SFA CP.

### benchmark:
- TTFT increased by **527%**
- TPOT increased by **180%**

<img width="1550" height="938" alt="image"
src="https://github.com/user-attachments/assets/9b7a03d8-a3db-4a99-8923-6e5bfcfecf72"
/>


### Does this PR introduce _any_ user-facing change?
None
### How was this patch tested?

- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef

---------

Signed-off-by: zzhx1 <zzh_201018@outlook.com>
Signed-off-by: zzhxx <zhangzihang23@mails.ucas.ac.cn>
Co-authored-by: clrs97 <524936896@qq.com>

vllm_ascend/ops/linear_op.py

@@ -62,7 +62,7 @@ from vllm_ascend.distributed.parallel_state import (get_flashcomm2_odp_group,
                                                     get_mlp_tp_group,
                                                     get_otp_group)
 from vllm_ascend.ops.flashcomm2_oshard_manager import flashcomm2_oshard_manager
-from vllm_ascend.utils import (enable_dsa_cp, enable_sp, flashcomm2_enable,
+from vllm_ascend.utils import (enable_dsa_cp, enable_dsa_cp_with_layer_shard, enable_sp, flashcomm2_enable,
                                get_flashcomm2_reorgnized_batch_ids,
                                matmul_allreduce_enable, mlp_tp_enable,
                                oproj_tp_enable, shared_expert_dp_enabled)
@@ -575,7 +575,8 @@ class SequenceRowParallelOp(CustomRowParallelOp):
             return tensor_model_parallel_all_reduce(output_parallel)
 
         pad_size = forward_context.pad_size
-        if pad_size > 0:
+        if pad_size > 0 and not (enable_dsa_cp()
+                                 and "o_proj" in self.layer.prefix):
             x = F.pad(x, (0, 0, 0, pad_size))
 
         world_size = self.layer.tp_size
@@ -728,7 +729,7 @@ def _get_row_parallel_op(
 ) -> Optional[Union[MLPRowParallelOp, OProjRowParallelOp,
                     Flashcomm2OProjRowParallelOp, MatmulAllreduceRowParallelOp,
                     SequenceRowParallelOp, ShardedCPRowParallelOp]]:
-    if enable_dsa_cp() and "o_proj" in prefix:
+    if enable_dsa_cp_with_layer_shard() and "o_proj" in prefix:
         return ShardedCPRowParallelOp(layer)
     if "down_proj" in prefix and mlp_tp_enable() and not is_moe_layer(prefix):
         return MLPRowParallelOp(layer)