[DLPack] Remove stride normalization when convert to DLPack

[SigureMo]

PR Category

Execute Infrastructure

PR Types

Bug fixes

Description

strides normalization 之后，原本 shape=[1, 2, 3, 4]、strides=[24, 12, 4, 1] 会变成 strides=[1, 12, 4, 1]，会在某些场景出现问题，因此移除，PyTorch 在如下 PR 后也已经移除

• [DLPACK] Optimize toDLPack Conversion Speed pytorch/pytorch#162111
• [Fix] Restrict stride normalization to 1D tensors on export pytorch/pytorch#163282
• [Fix] Completely remove stride normalization on DLPack Tensor pytorch/pytorch#164161

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

[Copilot]

Pull request overview

This PR removes stride normalization when converting Paddle tensors to DLPack format, addressing bugs that occur when strides are artificially normalized (e.g., setting stride=1 for dimensions with size < 2). The change aligns with PyTorch's recent similar modifications.

Key Changes:

• Removed stride normalization logic that modified strides for dimensions with size < 2
• Simplified memory management by using direct pointers to internal tensor metadata instead of heap-allocated copies
• Updated the deleter function to remove cleanup of shape/strides arrays (no longer heap-allocated)

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (develop@93ba43d). Learn more about missing BASE report.

Additional details and impacted files

@@             Coverage Diff             @@
##             develop    #77052   +/-   ##
===========================================
  Coverage           ?   100.00%           
===========================================
  Files              ?         1           
  Lines              ?         5           
  Branches           ?         0           
===========================================
  Hits               ?         5           
  Misses             ?         0           
  Partials           ?         0           

☔ 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.