Evening Robustness

TODO:

• [ ] Add page-specific notes for model-evaluation/field-evaluation/evening-robustness.md after inspecting the real project files.
• [ ] Cross-link Evening Robustness to the most relevant code, data, testing, and safety pages.
• [ ] Define the metric or field result in one precise sentence.
• [ ] List the script, command, or notebook that produces this result.
• [ ] List the exact dataset split, label file, and model versions being compared.
• [ ] Mark whether results are current or historical.
• [ ] Add units, thresholds, and interpretation rules.
• [ ] Add how this result can disagree with real field behavior.
• [ ] Add the retest checklist needed after labels or runtime code changes.
• [ ] Add route, lighting, weather, operator, model version, and manual takeovers.
• [ ] Separate subjective field notes from measurable field results.
• [ ] List model file path, version, preprocessing path, output scale, and known field behavior.
• [ ] Add current vs historical metrics once retesting is complete.
• [ ] Add test date, time, weather, route, model version, and manual takeover count.
• [ ] Add what failed, what worked, and what data should be collected next.