Returns Redis reachability + per-scraper freshness from scraper_runs.

Overall status is ok only when Redis is up, every required scraper's latest successful run is within 2 * expected_interval_s, every enabled canonical metal bar stream is fresh, and its writer owns the current lease. Otherwise status is degraded. Dead-man thresholds are read from scraper_runs rather than hardcoded constants (C1).

The value_staleness block (F1 per audit 2026-04-23) surfaces live FX rates that have stopped moving while the scraper keeps fetching. Per scope §3.2 Q2, entries in this block are INFORMATIONAL-ONLY — they do not flip overall status to degraded, so one stuck currency cannot leave /v1/status permanently degraded and drown out other alerts. Monitoring consumers alert on the presence of entries directly.

status and checks are computed at most once per _STATUS_REUSE_S per process and shared by concurrent callers; checks_at says when.