Update vllm pin to 12.24

[leo-pony]

What this PR does / why we need it?

Fix vllm break in the pr:

1. [Add MiMo-V2-Flash support] ([Model] Add MiMo-V2-Flash support vllm#30836)

Does this PR introduce any user-facing change?

How was this patch tested?

Co-authored-by: zxwang 1476209578@qq.com

• vLLM version: release/v0.13.0
• vLLM main: vllm-project/vllm@5fbfa8d

[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 updates the vLLM pin and introduces a v_head_size parameter to AscendQKVParallelLinear. However, the new parameter is not used in the layer's output size calculations, which is a critical bug. I've left a comment with details on how to fix it. The documentation change looks fine.

[leo-pony]

@shenchuxiaofugui Could you help take a look?
failed test case:
tests/e2e/multicard/test_qwen3_moe.py::test_qwen3_moe_w8a8_distributed_tp2_ep_dynamic_eplb FAILED

It's not a function issue. It's my local env issue.

[shenchuxiaofugui]

export HCCL_BUFFSIZE=1024
export DYNAMIC_EPLB="true"

vllm serve Qwen3-30B-A3B-m4
--served-model-name qwen3
--host 0.0.0.0
--port 20002
--tensor-parallel-size 2
--max_model_len 8192
--enable_expert_parallel
--quantization "ascend"
--enforce_eager
--additional-config '{"dynamic_eplb": true, "num_iterations_eplb_update": 100, "num_wait_worker_iterations": 50}'