Move checks from nonzero kernel to operator

[Silv3S]

• Early return if input to nonzero has no elements,
• Change pytorch MAX_DIMS to XPU_MAX_TENSORINFO_DIMS xpu equivalent,
• Rename tensor to out to match op schema.

[Copilot]

Pull Request Overview

This PR optimizes the nonzero operation by moving early validation checks from the kernel implementation to the operator level, improving performance for edge cases. The changes also standardize naming conventions and XPU-specific constants.

• Move early return for empty tensors from kernel to operator level
• Replace PyTorch MAX_DIMS with XPU-specific XPU_MAX_TENSORINFO_DIMS constant
• Rename tensor parameter to out for consistency with operation schema

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
src/ATen/native/xpu/Nonzero.cpp	Adds early return for empty tensors and updates dimension limit constant
src/ATen/native/xpu/sycl/NonzeroKernel.cpp	Removes redundant empty tensor check and renames variables for schema consistency

Comments suppressed due to low confidence (1)

src/ATen/native/xpu/sycl/NonzeroKernel.cpp:1

• The range tensor is declared but never used. The range_begin pointer is set to nullptr on line 99, but the range tensor allocation should be removed since it's not needed.

#include <ATen/Dispatch.h>

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}