[Optim][Cherry-pick] Reduce preemption occurrence when blocks not enough(#5696)

[rainyfly]

Motivation

当解码 block不足时，调度会抢占正在解码的请求，释放对应的 block 资源。并分配给剩余的解码请求。
之前的调度逻辑，在 waiting 队列里有请求的时候，发现剩余的 block可以容纳下一条新请求部分的 chunk（new_token_num)，就会将其调度回去做 prefill。在 block已经严重不足的时候，会造成反复 抢占->调度 prefill->抢占->调度 prefill的重调度行为，造成性能下降。
为了解决这一问题，在调度新请求做 prefill 时，考虑给正在解码的请求预留部分 block，只有在给正在解码的 每条请求所预留的 block剔除后，并且剩余的 block 还可以容纳整条当前需要 prefill 的请求，才把请求从 waiting 队列里调度出来做 prefill。

Modifications

新增环境变量：FD_RESERVE_OUTPUT_BLOCK_NUM_FOR_DECODE_WHEN_SCHEDULE_NEW_PREFILL
含义：从 waiting 队列里调度新请求做 prefill 时，需要给每条正在解码的请求预留的 block 数量，默认为 16

Usage or Command

v1 下默认使用

Accuracy Tests

None

Checklist

• Add at least a tag in the PR title.
  • Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
  • You can add new tags based on the PR content, but the semantics must be clear.
• Format your code, run pre-commit before commit.
• Add unit tests. Please write the reason in this PR if no unit tests.
• Provide accuracy results.
• If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

[paddle-bot]

Thanks for your contribution!

[codecov-commenter]

Codecov Report

❌ Patch coverage is 29.16667% with 17 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (release/2.4@fb59f56). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
fastdeploy/engine/sched/resource_manager_v1.py	29.16%	17 Missing ⚠️

Additional details and impacted files

@@              Coverage Diff               @@
##             release/2.4    #5808   +/-   ##
==============================================
  Coverage               ?   58.81%           
==============================================
  Files                  ?      329           
  Lines                  ?    40836           
  Branches               ?     6221           
==============================================
  Hits                   ?    24019           
  Misses                 ?    14945           
  Partials               ?     1872           

Flag	Coverage Δ
GPU	58.81% <29.16%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull request overview

这是一个从#5696 cherry-pick过来的优化PR，旨在减少在block资源不足时的抢占发生频率。该PR引入了一个预留块（reserve block）机制，在调度新的prefill请求时，会为正在解码的请求预留部分block资源，避免反复发生"抢占->调度prefill->抢占"的重调度行为。

主要变更：

• 添加了三个新的环境变量来控制预留块机制，包括初始预留数、衰减率和最小预留数
• 修改了调度逻辑，在检查是否可以调度新的prefill请求时，会考虑为当前运行中的decode请求预留block
• 实现了预留块数量的衰减机制，在正常调度时逐渐减少预留，在发生抢占时重置为初始值

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 9 comments.

File	Description
fastdeploy/envs.py	添加了三个新的环境变量用于配置预留块机制的参数（初始值、衰减率、最小值）
fastdeploy/engine/sched/resource_manager_v1.py	在ResourceManagerV1的初始化中添加了预留块相关的实例变量；修改了schedule()方法中的调度逻辑，在检查是否可以调度新prefill时计入预留块；在_trigger_preempt()中添加了预留块重置逻辑；在每次schedule()结束时实现预留块的衰减逻辑