Add more tests for batch invariant kernel-override logic [3/n]

[bwasti]

Following #25769, This change adds additional unit tests for the batch invariant kernel-overrides.

Purpose

We want more rigorous testing across kernels to ensure batch invariance. This PR lays out a test set to that end.

Test Plan

VLLM_KERNEL_OVERRIDE_BATCH_INVARIANT=1 pytest -s -v tests/v1/generation/batch_invariance/test_multi_gpu_ops.py

Test Result

Pass

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.
• (Optional) Release notes update. If your change is user facing, please update the release notes draft in the Google Doc.

[gemini-code-assist]

Code Review

This pull request significantly enhances the testing for batch invariance, particularly in multi-GPU scenarios, which is crucial for ensuring deterministic and correct model outputs. The introduction of test_multi_gpu_ops.py with its "needle-in-a-haystack" methodology provides a robust framework for verifying the behavior of parallel operators. The updates to existing tests to include randomized batch sizes and prompt lengths are also a welcome improvement. The necessary changes to the FlashInfer backend to support this batch-invariant mode appear correct and are well-contained. I have a couple of suggestions to improve code clarity and correctness.