[fast-ut] [reduced-it] [SkipRecovery] Restore randomized window aggregation coverage [databricks]

[wjxiz1992]

JaCoCo sql-plugin line coverage: +0 lines (shim 350 vs nightly b9; the raw +2 lines were the excluded recurring MemoryCheckerImpl local GPU-initialization artifact)

Contributes to #9682.

Summary

• Remove the stale datagen_overrides(seed=0) marker from test_window_aggs_for_rows.
• Restore randomized data generation after Adopt changes from JNI for casting from float to decimal #10917 fixed the float-to-decimal conversion path that closed [BUG] Casting FLOAT64 to DECIMAL(12,7) produces different rows from Apache Spark CPU #9682.
• Keep the test body, eight data generators, two batch sizes, configs, and CPU/GPU equality assertion unchanged.

Validation

• Spark 3.5 build: BUILD SUCCESS; 11/11 reactor modules succeeded.
• Original failure seed before marker removal: 2 passed, 1038 deselected, 4 warnings in 8.50s.
• Marker-free focused matrix at seeds 1698940723, 0, 1, and 42: 64/64 parameter instances passed.
• Post-rebase original-seed gate: 16 passed, 4 warnings in 26.49s.
• Full window_function_test.py: 1031 passed, 5 skipped, 4 xfailed, 535 warnings in 800.29s.
• GPU event log contains GpuWindow, GpuProject, GpuSort, and GpuShuffleExchangeExecBase.
• NT local review: 6 independent reviewers, 0 findings.

Documentation

• Updated for new or modified user-facing features or behaviors
• No user-facing change

Testing

• Added or modified tests to cover new code paths
• Covered by existing tests
  (Please provide the names of the existing tests in the PR description.)
• Not required

Performance

• Tests ran and results are added in the PR description
• Issue filed with a link in the PR description
• Not required

[Copilot]

Pull request overview

Restores randomized coverage for test_window_aggs_for_rows in the Python integration test suite by removing a temporary seed override that was added to avoid failures related to float-to-decimal casting differences (now fixed upstream).

Changes:

• Removes the stale @datagen_overrides(seed=0, ...) marker from test_window_aggs_for_rows, re-enabling normal randomized datagen behavior for this test.
• Leaves the test logic, parametrization matrix, and CPU/GPU equality assertions unchanged.

[greptile-apps]

Greptile Summary

Restores randomized-seed coverage for test_window_aggs_for_rows by removing its stale seed override while leaving the test matrix, query, configuration, and CPU/GPU equality assertion unchanged.

Confidence Score: 5/5

The PR appears safe to merge because the marker removal allows configured seed variation without changing the tested query or comparison behavior.

The integration-test harness supplies the same selected seed to both CPU and GPU data generation, and removing this non-permanent seed override introduces no unsupported execution or reproducibility failure.

Important Files Changed

Filename	Overview
integration_tests/src/main/python/window_function_test.py	Removes the obsolete datagen_overrides(seed=0) marker from the rows-based window aggregation integration test; no actionable defect was identified.

_{Reviews (1): Last reviewed commit: "test: restore randomized window aggregat..." | Re-trigger Greptile}

[wjxiz1992]

build