[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/spec_decode/mtp_proposer.py

@@ -60,6 +60,8 @@ class MtpProposer(Proposer):
         self.torchair_compiled_models = {}  # type: ignore
         self.torchair_graph_enabled = get_ascend_config(
         ).torchair_graph_config.enabled
+        self.enable_shared_expert_dp = get_ascend_config(
+        ).enable_shared_expert_dp
         # We need +1 here because the arange is used to set query_start_loc,
         # which has one more element than batch_size.
         self.arange = torch.arange(vllm_config.scheduler_config.max_num_seqs +
@@ -79,7 +81,9 @@ class MtpProposer(Proposer):
         with set_default_torch_dtype(
                 draft_model_config.dtype), set_current_vllm_config(
                     self.vllm_config):
-            if self.torchair_graph_enabled:
+            if self.torchair_graph_enabled or (
+                    self.enable_shared_expert_dp
+                    and self.vllm_config.model_config.use_mla):
                 self.model = TorchairDeepSeekMTP(
                     vllm_config=self.vllm_config).to(target_device)
             else: