[XPU] Add forward_xpu to Gemma4RMSNorm by reusing existing kernel ops

[CaoE]

Motivation

Add forward_xpu to Gemma4RMSNorm by reusing ops already available in sgl-kernel for XPU.

Modifications

Refactor forward_cuda into a shared _forward_impl helper with CUDA path remaining unchanged.
Add forward_xpu that leverages the XPU kernel's native 3D stride support.
Note: The XPU gemma_rmsnorm/ rmsnorm kernels already handle non-contiguous 3D inputs via stride-aware row offset computation, so we can skip the flatten-to-2D reshape when the tensor dim is 3 with contiguous last dim.

Generated by my agent.

Accuracy Tests

Covered by sgl-project/sgl-kernel-xpu#216.

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review and Merge Process

1. Ping Merge Oncalls to start the process. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • Common commands include /tag-and-rerun-ci, /tag-run-ci-label, /rerun-failed-ci
4. After green CI and required approvals, ask Merge Oncalls or people with Write permission to merge the PR.

---

CI States

Latest PR Test (Base): 🚫 Run #26428790998
Latest PR Test (Extra): ❌ Run #26428790936

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[CaoE]

/tag-run-ci-label

[CaoE]

This is coverd by #23280.