feat: sync runtime search and schema quality updates from app repo

- port retrieval, validation, and eval improvements relevant to os
- align prompts and dimensions with the flat single-agent model
- replace the old eval suite with the focused core scenarios

Generated with Codex

tests/evals/scenarios/archive/youtube-extract.ts

@@ -0,0 +1,16 @@
+import { Scenario } from '../types';
+
+export const scenario: Scenario = {
+  id: 'youtube-extract',
+  name: 'YouTube extract (optional)',
+  description: 'Extract and ingest a YouTube video (requires network + API keys).',
+  tools: ['youtubeExtract'],
+  enabled: false,
+  notes: 'Enable when network + API keys are configured; can be flaky.',
+  input: {
+    message: 'Add this video: https://www.youtube.com/watch?v=dQw4w9WgXcQ',
+  },
+  expect: {
+    toolsCalledSoft: ['youtubeExtract'],
+  },
+};