[Feature] Support to use fullgraph with eagle (#5118)

### What this PR does / why we need it?
    
We support to use full graph with eagle. 

Change list:
1. Distinguish between processing graph_params and draft_graph_params in
attention_v1.
    2. Adapt the full-graph mode in eagle_proposer, include:
        1). If use full graph, make Fullgraph Wrapper when load model.
2). Build a new meatadata, set running mode in FULL and mark attention
update in dummy_run when in Fullgraph mode.
3). Fixed and fill any attn_metadata, such as
attn_metadata.slot_mapping.
        4). Add a descriptor.
        5). Set running mode and triggered update metadata.
3. Trans is_mtp_model to is_draft_model, and add the update of
workspace.

NOTE:
When set async_scheduling=True, the draft model will enforce execution
in eager mode.

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

---------

Signed-off-by: anon189Ty <Stari_Falcon@outlook.com>
Co-authored-by: Yizhou Liu <liu_yizhou@outlook.com>
Co-authored-by: Yizhou <136800916+yiz-liu@users.noreply.github.com>

vllm_ascend/ascend_forward_context.py

@@ -36,7 +36,7 @@ def set_ascend_forward_context(
         aclgraph_runtime_mode: CUDAGraphMode = CUDAGraphMode.NONE,
         batch_descriptor: Optional[BatchDescriptor] = None,
         model_instance: torch.nn.Module = None,
-        is_mtp_model=False):
+        is_draft_model=False):
     """A context manager that stores the current forward context,
     can be attention metadata, etc.
     We add some additional param into forward_context.
@@ -55,7 +55,7 @@ def set_ascend_forward_context(
         from vllm_ascend.ops.fused_moe.moe_comm_method import \
             get_moe_comm_method
         moe_comm_type = select_moe_comm_method(num_tokens, vllm_config,
-                                               is_mtp_model)
+                                               is_draft_model)
         forward_context.moe_comm_type = moe_comm_type
         forward_context.moe_comm_method = get_moe_comm_method(moe_comm_type)
 
@@ -110,7 +110,7 @@ def set_ascend_forward_context(
             forward_context.prefetch_mlp_down_proj = False
         forward_context.prefetch_mlp_enabled = prefetch_mlp_enabled
         forward_context.model_instance = model_instance
-        forward_context.is_mtp_model = is_mtp_model
+        forward_context.is_draft_model = is_draft_model
 
         if num_tokens is None and attn_metadata is not None:
             num_tokens = attn_metadata.num_actual_tokens
@@ -195,7 +195,7 @@ def get_mc2_mask():
 
 def select_moe_comm_method(num_tokens: int,
                            vllm_config: VllmConfig,
-                           is_mtp_model=False) -> Optional[MoECommType]:
+                           is_draft_model=False) -> Optional[MoECommType]:
     """Select the MoE communication method according to parallel settings,
     device generation, token count, and quantization.
 
@@ -210,7 +210,7 @@ def select_moe_comm_method(num_tokens: int,
     Args:
         num_tokens (int): The number of tokens in the current batch.
         vllm_config (VllmConfig): Runtime configuration for the model.
-        is_mtp_model (bool): Whether the model runs in MTP mode (disables fused MC2).
+        is_draft_model (bool): Whether the model runs in MTP mode (disables fused MC2).
 
     Raises:
         ValueError: If the soc version is unsupported.
@@ -249,13 +249,13 @@ def select_moe_comm_method(num_tokens: int,
             fused_decode_enable = fused_mc2_enable
             if envs_ascend.VLLM_ASCEND_ENABLE_FUSED_MC2 == 1:
                 fused_decode_enable = fused_mc2_enable and get_ep_group(
-                ).world_size <= 16 and (not is_mtp_model)
+                ).world_size <= 16 and (not is_draft_model)
             moe_comm_type = MoECommType.FUSED_MC2 if fused_decode_enable else MoECommType.MC2
         else:
             fused_prefill_enable = fused_mc2_enable
             if envs_ascend.VLLM_ASCEND_ENABLE_FUSED_MC2 == 1:
                 fused_prefill_enable = fused_mc2_enable and get_ep_group(
-                ).world_size <= 16 and (not is_mtp_model)
+                ).world_size <= 16 and (not is_draft_model)
             elif envs_ascend.VLLM_ASCEND_ENABLE_FUSED_MC2 == 2:
                 fused_prefill_enable = False
             moe_comm_type = MoECommType.FUSED_MC2 if fused_prefill_enable else MoECommType.ALLTOALL