[long_seq_Feat] support chunk prefill

[LookAround0301]

What this PR does / why we need it?

1、qwen GQA attention_v1 optim
2、DeepSeek MLA refactor, all gather q -> all gather kv
3、modelrunner refactor for chunk prefill, we remove some code not use

Does this PR introduce any user-facing change?

How was this patch tested?

• vLLM version: v0.11.0
• vLLM main: vllm-project/vllm@2918c1b

[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 refactors the chunked prefill implementation for MLA attention to better support context parallelism (PCP/DCP). The changes simplify the logic in _compute_prefill_context by introducing a new helper function _reorg_kvcache and unifying the handling of different parallelism configurations. While the refactoring in vllm_ascend/attention/mla_v1.py is a significant improvement in code clarity and maintainability, I've identified a critical bug in vllm_ascend/worker/model_runner_v1.py related to the calculation of context length for speculative decoding when context parallelism is enabled. This could lead to incorrect attention computations.

[wangxiyuan]

I'll enable full test once lint pass