[Main] [Refactor] Enable MoECommMethod in Eager Mode (#2791)

### What this PR does / why we need it?
1. Replace prepare/finalize operation in fused_moe.py by
moe_comm_method.prepare()/finalize()
2. Replace unified_fused_experts by moe_comm_method.fused_experts() in
fused_moe.py/w8a8_dynamic.py/w4a8_dynamic.py
3. Add calling _select_moe_comm_method in spec-decode proposers.
4. Currently, w4a8_dynamic does not support gatherep, use all2allv
instead.
5. Remove redundant code.
### Does this PR introduce _any_ user-facing change?
AllgatherEP switch is disabled in aclgraph/eager mode, just follow the
rules in modelrunner_v1._select_moe_comm_method()
### How was this patch tested?
e2e & ut


- vLLM version: v0.10.2
- vLLM main:
7f6f2c1182

Signed-off-by: Pr0Wh1teGivee <calvin_zhu0210@outlook.com>
Co-authored-by: weijinqian0 <12153182+weijinqian0@users.noreply.github.com>

tests/ut/models/test_deepseek_v2.py

@@ -23,8 +23,7 @@ from vllm.distributed.parallel_state import GroupCoordinator
 
 from vllm_ascend.models.deepseek_v2 import (
     CustomDeepseekV2MergedReplicatedLinear, CustomDeepseekV2MLAAttention,
-    CustomDeepseekV2MLP, CustomDeepseekV2MoE,
-    CustomDeepseekV2RowParallelLinear,
+    CustomDeepseekV2MLP, CustomDeepseekV2RowParallelLinear,
     CustomDeepseekV2RowParallelLinearReplaceAllreduce,
     CustomDeepseekV2SiluAndMul, LogitsProcessor, ParallelLMHead)
 
@@ -213,22 +212,6 @@ def test_custom_deepseek_v2_mlp(mock_distributed, base_config):
                             quant_config=None)
 
 
-def test_custom_deepseek_v2_moe(mock_distributed, base_config,
-                                mock_forward_context):
-    base_config.n_shared_experts = 1
-    moe = CustomDeepseekV2MoE(config=base_config,
-                              quant_config=None,
-                              prefix="mlp")
-    assert moe.top_k == 2
-
-    x = torch.randn(2, 4, 128)
-    attn_metadata = Mock(num_prefills=1)
-    with patch("vllm_ascend.ops.fused_moe.AscendFusedMoE.__call__",
-               return_value=(torch.randn(2, 4, 128), torch.randn(2, 4, 128))):
-        output = moe(x, attn_metadata)
-        assert output.shape == (2, 4, 128)
-
-
 @patch("torch_npu.npu_rms_norm")
 def test_custom_deepseek_v2_mla_attention(mock_rms_norm, mock_distributed,
                                           base_config):