[Speculative Decoding] fix mtp stop_seqs and limit thinking bugs

[lonelygsh]

Motivation

本 PR 修复投机解码中 speculate_set_stop_value_multi_seqs 和 speculate_limit_thinking_content_length 两个 kernel 的多个问题。

Modifications

speculate_set_stop_value_multi_seqs

1. stop_seqs 不对比最后一个 accept token：将循环条件从 accept_idx <= accept_num - 1 改为 accept_idx < accept_num - 1。
2. 加入回退逻辑：匹配到 stop sequence 后，丢弃其后的多余 token，并将 step_idx 回退相应步数，确保状态一致。
3. 修复截断逻辑：匹配到 stop sequence 后，保留完整的 stop sequence，在其后追加 EOS token（对齐非 MTP 行为：...<|im_end|> <eos>）。
4. 修复索引计算：修正 accept_tokens 和 pre_ids 的边界判断及索引计算。

speculate_limit_thinking_content_length

5. 修复截断判断条件：将 current_step - 1 == max_think_len 改为 current_step == max_think_len，修正 off-by-one 错误。

测试

• 更新 test_speculate_set_stop_value_multi_seqs.py，同步适配新的 kernel 语义，新增 step_idx 回退验证。

Usage or Command

无新增接口，修复已有逻辑。可通过投机解码推理验证 stop sequences 截断行为及 thinking 长度限制是否正确。

Accuracy Tests

单元测试通过。

Checklist

• Add at least a tag in the PR title.
• Format your code, run pre-commit before commit.
• Add unit tests. 已更新 test_speculate_set_stop_value_multi_seqs.py。
• 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!

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

guanshihui] seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (develop@98f3fc9). Learn more about missing BASE report.

Additional details and impacted files

@@            Coverage Diff             @@
##             develop    #7166   +/-   ##
==========================================
  Coverage           ?   73.90%           
==========================================
  Files              ?      376           
  Lines              ?    52867           
  Branches           ?     8243           
==========================================
  Hits               ?    39072           
  Misses             ?    11071           
  Partials           ?     2724           

Flag	Coverage Δ
GPU	73.90% <ø> (?)

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.