[Refactor] Adjustments to moe_comm_method selection process (#3001)

### What this PR does / why we need it?
Fix issues mentioned in
https://github.com/vllm-project/vllm-ascend/pull/2791 and some minor
refactoring.
1. Use Enum instead of string.
2. Avoid setting a new property to forward_context in
AscendFusedMoE.forward().
3. Enabling TokenDispatcherWithMoge.
4. Remove redundant code.

### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?

Qwen3-30B-A3B/Qwen3-30B-A3B-W8A8/DeepSeek-V3-W4A8-Pruing/deepseek-mtp/pangu-pro-moe-pruing:
1. Enable/Disable EP
2. Aclgraph & eager


- vLLM version: v0.10.2
- vLLM main:
9607d5eb44

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

vllm_ascend/spec_decode/eagle_proposer.py

@@ -117,11 +117,11 @@ class EagleProposer(Proposer):
                   skip_attn: bool = False,
                   num_reqs: int = 0,
                   num_tokens_across_dp: Optional[torch.Tensor] = None):
-        moe_comm_method = self.runner._select_moe_comm_method(
+        moe_comm_type = self.runner._select_moe_comm_method(
             num_tokens, with_prefill)
         with set_ascend_forward_context(None,
                                         self.vllm_config,
-                                        moe_comm_method=moe_comm_method,
+                                        moe_comm_type=moe_comm_type,
                                         num_tokens=num_tokens):
             self.model(
                 input_ids=self.input_ids[:num_tokens],
@@ -454,7 +454,7 @@ class EagleProposer(Proposer):
         with_prefill = attn_metadata.attn_state not in [
             AscendAttentionState.DecodeOnly, AscendAttentionState.SpecDecoding
         ]
-        moe_comm_method = self.runner._select_moe_comm_method(
+        moe_comm_type = self.runner._select_moe_comm_method(
             num_input_tokens, with_prefill)
 
         # copy inputs to buffer for cudagraph
@@ -463,7 +463,7 @@ class EagleProposer(Proposer):
         attn_metadata.block_tables = block_table.to(device)
         with set_ascend_forward_context(attn_metadata,
                                         self.vllm_config,
-                                        moe_comm_method=moe_comm_method,
+                                        moe_comm_type=moe_comm_type,
                                         num_tokens=num_input_tokens):
             last_hidden_states, hidden_states = self.model(
                 input_ids=self.input_ids[:num_input_tokens],
@@ -495,7 +495,7 @@ class EagleProposer(Proposer):
         else:
             input_batch_size = batch_size
 
-        moe_comm_method = self.runner._select_moe_comm_method(
+        moe_comm_type = self.runner._select_moe_comm_method(
             input_batch_size, False)
 
         attn_metadata.num_actual_tokens = batch_size
@@ -568,7 +568,7 @@ class EagleProposer(Proposer):
             # Run the model.
             with set_ascend_forward_context(attn_metadata,
                                             self.vllm_config,
-                                            moe_comm_method=moe_comm_method,
+                                            moe_comm_type=moe_comm_type,
                                             num_tokens=input_batch_size):
 
                 last_hidden_states, hidden_states = self.model(