[Refactor] Unify full-graph parameter update logic

[LICO1314]

What this PR does / why we need it?

Refactor: Unify full-graph parameter update logic

This PR consolidates the scattered full-graph parameter update logic into a unified approach, improving code architecture and eliminating duplication.

Key improvements:

1. Unified interface

   • Create update_full_graph_params as the single entry point for all full-graph updates
   • Replace multiple scattered update calls with one unified function
   • Remove ~50 lines of duplicated if-else logic across model_runner_v1.py and eagle_proposer.py

2. Better architecture

   • Move update logic to respective Backend classes (AscendAttentionBackend, AscendMLABackend)
   • Each Backend manages its own parameter update logic internally
   • Simplify caller code to just dispatch to the appropriate Backend

3. Cleaner parameter handling

   • Remove unnecessary pcp_size and dcp_size parameter passing
   • Get parallel configuration directly from distributed groups
   • Consistent with how other parts of the codebase obtain these values

Why we need it:

• Maintainability: Future changes only need to be made in one place per Backend
• Code quality: Follows DRY principle and Single Responsibility Principle
• Readability: Cleaner, more intuitive code structure

Does this PR introduce any user-facing change?

No. This is a pure refactoring with no functional changes - same behavior, cleaner code.

How was this patch tested?

• All existing unit tests pass with updated mocks
• No new tests needed (pure refactoring, no behavior changes)
• CI validates correctness

---

• vLLM version: v0.13.0

[github-actions]

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

• A PR should do only one thing, smaller PRs enable faster reviews.
• Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
• Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

[gemini-code-assist]

Code Review

This pull request introduces two significant improvements: a refactoring of the attention parameter update logic into a unified function update_full_graph_params, and the addition of full-graph support for Qwen3-Next-MTP, which should improve performance. The refactoring simplifies the codebase by removing duplicated logic, which is a great enhancement for maintainability. The changes to enable full-graph support for MTP are also well-implemented. I've found one critical issue where .contiguous() calls were omitted when inlining a helper function, which could lead to runtime errors. My feedback includes a specific code suggestion to fix this.

[gemini-code-assist]

The .contiguous() calls are missing for positions and previous_hidden_states before passing them to torch.ops.vllm.maybe_all_gather_and_maybe_unpad. Since these tensors are created by slicing larger tensors (self.positions and self.hidden_states), they may not be contiguous. This could lead to runtime errors or incorrect behavior in the custom op, as the original maybe_all_gather_and_unpad helper function included these calls to ensure contiguity.

Suggested change

	positions = torch.ops.vllm.maybe_all_gather_and_maybe_unpad(
	positions, True)
	previous_hidden_states = torch.ops.vllm.maybe_all_gather_and_maybe_unpad(
	previous_hidden_states, True)
	positions = torch.ops.vllm.maybe_all_gather_and_maybe_unpad(
	positions.contiguous(), True)
	previous_hidden_states = torch.ops.vllm.maybe_all_gather_and_maybe_unpad(
	previous_hidden_states.contiguous(), True)

[github-actions]

This pull request has conflicts, please resolve those before we can evaluate the pull request.

[wangxiyuan]

CI is broken after this PR merged: https://github.com/vllm-project/vllm-ascend/actions/runs/21317195670/job/61361650122?pr=6226