feat: add pd prefill short request first scheduling.

[immengzi]

Description

Adds opt-in ShortRequestFirst scheduling for PD-prefill requests.

• Adds enable_short_request_first, short_request_first_threshold, and short_request_first_long_max_wait_ms through CLI flags and JSON configuration.
• Splits online PD-prefill requests into immediate, short, and long queues. Requests with partial KV state or preemption recovery are scheduled first; fresh short requests are then prioritized over fresh long requests.
• Supports optional long-request aging to prevent starvation. A value of 0 retains strict short-request-first behavior.
• Keeps offline requests in the existing queue and preserves online-request scheduling priority.
• Restricts the feature to FCFS PD-prefill/MIX scheduling with chunked prefill enabled.
• Adds queue-depth and aged-long-promotion metrics.
• Adds configuration, queue, and scheduler unit tests, together with PD startup and AISBench benchmark/replay helper scripts.

Related Issues

N/A

Change Type

• Bug fix
• New feature
• Performance improvement
• Refactor
• Documentation
• Test
• Build or CI

Pull Request Checklist

PR Title and Commit Messages

• The PR title and each commit message follow the xLLM commit format: <type>: <subject>.

Pre-commit Checks

• I have installed pre-commit by running pip install pre-commit or an equivalent command.
• I have installed the hooks with pre-commit install.
• I have run pre-commit run --all-files and fixed any reported issues.

Self Review

• I have self-reviewed the code according to .agents/skills/code-review/references/custom-code-style.md, especially code written or assisted by AI.
• I have rebased this PR onto the latest main branch.

Build and Test Coverage

• Tests have been added or updated as needed.
• CUDA: python setup.py build test has passed on a CUDA machine.
• NPU: python setup.py build test has passed on an NPU machine.
• MLU: python setup.py build test has passed on an MLU machine.

Reviewer Notes

No known functional issues.