[misc][torchair] fix bugs around deepseek mtp, enable_shared_expert_dp and use_cached_kv_cache_bytes (#3074)

### What this PR does / why we need it?
This miscellaneous​ contains several small fixes:
1) fix initialization and forward bugs of DeepseekMTPLayer with
`shared_expert_dp` enabled.
2) fix a tensor shape mismatches after o_proj caused by a work-aroud
change in NPUModelRunner.
3) avoid unnecessary decline of kv_cache memory (default: 64MB) with
`use_cached_kv_cache_bytes` disabled.
4) fall back `fused_moe_state` from `MC2` to `All2All` since the padding
logic of `mc2_mask` is incompatible with input hidden_states when
`shared_expert_dp` enabled.

Once this PR is merged, users can launch disaggregated_prefill
deployments (large_ep) with `deepseek_mtp` and `shared_expert_dp` as
`v0.9.1-dev` branch. The remaining problem of kv_cache tokens decline
compared to `v0.9.1-dev` will be resolved by
https://github.com/vllm-project/vllm-ascend/pull/3073.
 
### Does this PR introduce _any_ user-facing change?

No.
### How was this patch tested?
E2E vllm serving about deepseek_mtp with torchair graph mode and
`enable_shared_expert_dp` with eager mode. Large ep deployments are also
tested with this PR.


- vLLM version: v0.10.2
- vLLM main:
5aeb925452

---------

Signed-off-by: linfeng-yuan <1102311262@qq.com>

vllm_ascend/torchair/quantization/torchair_w8a8_dynamic.py

@@ -833,6 +833,7 @@ class TorchairAscendW8A8DynamicFusedMoEMethod:
 
         ascend_config = get_ascend_config()
         self.torchair_graph_enabled = ascend_config.torchair_graph_config.enabled
+        self.enable_shared_expert_dp = ascend_config.enable_shared_expert_dp
 
         try:
             device_group = get_mc2_group().device_group
@@ -946,6 +947,8 @@ class TorchairAscendW8A8DynamicFusedMoEMethod:
             )
 
         fused_moe_state = get_forward_context().fused_moe_state
+        if self.enable_shared_expert_dp and fused_moe_state == FusedMoEState.MC2:
+            fused_moe_state = FusedMoEState.All2All
         shared_gate_up, shared_dequant_scale = None, None
         if shared_experts is not None and fused_moe_state == FusedMoEState.MC2:
             with npu_stream_switch("moe_secondary", 0):