[Cherry-Pick] Add clear_grpah_opt_backend method to Qwen3VL(#7086)

[Guo-Yilong]

Motivation

Cherry-pick from develop branch.

Add clear_grpah_opt_backend method to Qwen3VLForConditionalGeneration to support clearing cuda graph optimization backend, enabling proper cleanup when switching inference backends.

Modifications

• Add clear_grpah_opt_backend method in Qwen3VLForConditionalGeneration that delegates to the underlying model to clear cuda graph optimization backend.

Usage or Command

N/A

Accuracy Tests

No model output changes involved.

Checklist

• Add unit tests. 仅新增一个委托方法，暂不添加单测。
• Documentation(README/CNAME/Doxyfile) modified or not. No
• Usage/AutoTest modified or not. No
• Accuracy test modified or not. No
• FastDeploy library interface modified or not. No
• Other modified or not. No

[paddle-bot]

Thanks for your contribution!

[codecov-commenter]

Codecov Report

❌ Patch coverage is 50.00000% with 1 line in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (release/2.5@054add2). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
...tdeploy/model_executor/models/qwen3_vl/qwen3_vl.py	50.00%	1 Missing ⚠️

Additional details and impacted files

@@              Coverage Diff               @@
##             release/2.5    #7110   +/-   ##
==============================================
  Coverage               ?   68.42%           
==============================================
  Files                  ?      390           
  Lines                  ?    54091           
  Branches               ?     8515           
==============================================
  Hits                   ?    37012           
  Misses                 ?    14393           
  Partials               ?     2686           

Flag	Coverage Δ
GPU	68.42% <50.00%> (?)

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.

[fastdeploy-bot]

🤖 AI Code Review | {{TIMESTAMP}}

📋 Review 摘要

PR 概述：Cherry-pick 为 Qwen3VLForConditionalGeneration 类添加 clear_grpah_opt_backend 方法，支持清理 cuda graph 优化后端
变更范围：model_executor/models/qwen3_vl/
影响面 Tag：[Models] [Graph Optimization]

📝 PR 规范检查

标题使用 [Cherry-Pick] 不在官方 Tag 列表中，建议使用标准 Tag。

标题建议（可直接复制）：

• [Models] Add clear_grpah_opt_backend method to Qwen3VL (#7086 cherry-pick)

问题

未发现阻塞性问题。

总体评价

代码实现正确，遵循了项目中其他模型（如 qwen3.py、qwen2.py、deepseek_v3.py）相同的委托模式。方法名 clear_grpah_opt_backend 虽然存在拼写问题（grpah → graph），但与项目现有命名保持一致，是正确的做法。

[CSWYF3634076]

LGTM