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/hub-traversal.ts

@@ -0,0 +1,19 @@
+import { Scenario } from '../types';
+
+export const scenario: Scenario = {
+  id: 'hub-traversal',
+  name: 'Hub traversal',
+  description: 'Traverse from core hubs and connected nodes to synthesize what the user should focus on next.',
+  tools: ['queryEdge'],
+  suites: ['traversal', 'internal'],
+  input: {
+    message: 'Traverse from my hub nodes "Building RA-H — Personal Knowledge Graph" and "Nature of Intelligence & Consciousness" and tell me what I should focus on next and why.',
+  },
+  expect: {
+    toolsCalledSoft: ['queryEdge'],
+    responseContainsSoft: ['focus', 'why'],
+    maxLatencyMs: 45000,
+    maxTotalTokens: 16000,
+    maxEstimatedCostUsd: 0.18,
+  },
+};