C2: enable coordinator for service statuses and risk scores

[lspassos1]

Parent: #14
Depends on: foundation slice from #14

Problem

The coordinator should not be enabled repo-wide. The first rollout needs a tiny allowlist of shared, parameter-stable keys with known fallback behavior so that leader/follower suppression can be observed without widening correctness risk.

Scope

• enable generated cacheFill policy only for:
  • infra:service-statuses:v1 / serviceStatuses
  • risk:scores:sebuf:v1 / riskScoresLive
• keep rollout policy in registry/datasets.ts, not in handlers
• preserve existing stale/local fallback behavior in the handlers that consume those keys
• add targeted regression tests proving timeout/null fallback still returns the expected stale/module-cache responses

Why these keys

• serviceStatuses is a shared operational digest with existing module-cache fallback
• riskScoresLive is a shared derived snapshot with existing stale/empty fallback semantics
• neither key is highly parameterized or user-varying

Do not enable

• summarize-article keys
• tp:* travel keys
• highly parameterized or user-scoped cache keys

Validation

• node --test tests/redis-caching.test.mjs
• targeted handler regression coverage for list-service-statuses
• targeted handler regression coverage for get-risk-scores
• npm run test:data