[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/ops/torchair_fused_moe.py

@@ -803,6 +803,7 @@ class TorchairAscendUnquantizedFusedMoEMethod(UnquantizedFusedMoEMethod):
 
         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
@@ -884,6 +885,8 @@ class TorchairAscendUnquantizedFusedMoEMethod(UnquantizedFusedMoEMethod):
             topk_ids = torch.randint_like(topk_ids, 0, global_num_experts)
 
         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
 
         if fused_moe_state == FusedMoEState.MC2:
             return torchair_fused_experts_with_mc2(
@@ -1155,6 +1158,8 @@ class TorchairAscendFusedMoE(FusedMoE):
         forward_context = get_forward_context()
         fused_moe_state = forward_context.fused_moe_state
         mc2_mask = forward_context.mc2_mask
+        if self.enable_shared_expert_dp and fused_moe_state == FusedMoEState.MC2:
+            fused_moe_state = FusedMoEState.All2All
         # For w8a8 dynamic we can do npu_dynamic_quant and gate in parallel.
         quantized_x_for_share, dynamic_scale_for_share = None, None
         from vllm_ascend.torchair.quantization.torchair_w8a8_dynamic import \