Fix backward operator return values to respect requires_grad settings

[guozixu2001]

Motivation and Context

There are some error when running resnet50 testcase

This change addresses an issue with the conv2d_backward function where it was returning unnecessary gradients even when requires_grad was not set for some parameters. By respecting the requires_grad settings, we ensure that only the required gradients are computed and returned.

Description

• Modified the backward operator function to ensure that gradients are only included in the return dictionary if requires_grad is set to True for the respective parameters.
• Updated the test template to correctly handle and pass requires_grad information from function_paras.
• Updated the diopiTensor class definition to include the requires_grad attribute.

Use cases (Optional)

BC-breaking (Optional)

Checklist

Before PR:

• [x] I have read and followed the workflow indicated in the Contributors.md to create this PR.
• [x] Pre-commit or linting tools indicated in Contributors.md are used to fix the potential lint issues.
• [x] Bug fixes are covered by unit tests, the case that causes the bug should be added in the unit tests.
• [x] New functionalities are covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• [x] The documentation has been modified accordingly, including docstring or example tutorials.

After PR:

• [x] CLA has been signed and all committers have signed the CLA in this PR.