🤖 Update Benchmark Results - #5449

Benchmark Update Review

This is a straightforward automated benchmark data refresh for TUnit v1.29.0 (up from v1.28.7), with MSTest also bumped to v4.2.1.

Performance Summary

Runtime wins for TUnit v1.29.0:

• AsyncTests: 537.0ms → 528.7ms (~1.5% faster), AOT: 122.7ms → 120.9ms
• DataDrivenTests: 477.77ms → 473.79ms (~0.8% faster), AOT: 24.35ms → 22.25ms (~8.6% faster — notable AOT improvement)
• SetupTeardownTests: 582.0ms → 577.9ms (~0.7% faster)
• ScaleTests AOT: 36.53ms → 36.02ms (~1.4% faster)

Slight regressions (within noise):

• BuildTime: 1.789s → 1.842s (+3%); all frameworks moved up similarly (NUnit: +0.05s, xUnit3: +0.027s), suggesting CI environment variance rather than a real regression
• MassiveParallelTests: 655.8ms → 665.1ms (+1.4%); again, NUnit and xUnit3 also moved up, pointing to environmental noise
• ScaleTests: 445.59ms → 455.16ms (+2.1%); MSTest actually improved from 553ms to 526ms in the same run, so this is likely CI noise

Observation on DataDrivenTests: MSTest appears to have regressed significantly in this benchmark (475.82ms → 590.47ms, ~24% slower). That's a notable jump — though this is their version bump from 4.1.0 → 4.2.1, not a TUnit concern.

Structural Quality

The automated pipeline produces consistent JSON and Markdown output. All files (JSON data sources, rendered Markdown, historical trends) are updated atomically in a single PR, which is the right approach.

No issues found. The changes are correct, internally consistent, and the performance story for TUnit remains strong.