[bugfix]support dsv3.2 enable both mtp and full_decode_only

[cookieyyds]

What this PR does / why we need it?

#5230 this PR introduced a problem when both mtp and full_decode_only are enabled for the DSV32 model, the operators cannot be compiled into the graph. This PR fixes that issue.

• vLLM version: v0.13.0
• vLLM main: vllm-project/vllm@2f4e654

[gemini-code-assist]

Code Review

This pull request addresses a bug that prevented CUDA graph compilation for the DSV3.2 model when both sequence parallelism (MTP) and the full_decode_only graph mode are enabled. The fix correctly changes the CUDA graph support level reported by the AscendSFAMetadataBuilder from AttentionCGSupport.UNIFORM_SINGLE_TOKEN_DECODE to AttentionCGSupport.UNIFORM_BATCH. This change aligns the SFA attention backend with the MLA backend, enabling graph capture for a wider range of uniform batches. The modification is logical, directly addresses the described problem, and appears to be correct. I approve this change.

[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.