feat(insights): show the wellness assessment and add score explainers

The score-anatomy detail page now renders the per-score AI assessment that
already rides GET /api/insights/derived (shipped to iOS in v1.13.2) — the web
type dropped the field, so the card was silently discarded. The assessment
mounts below the composition card via the same InsightStatusCard the metric
sub-pages use, so it reads identically for readiness, sleep, recovery, strain
and stress.

Each score detail page also gains an educational intro caption beneath the
heading, mirroring the metric sub-pages: new explainer bodies for the five
scores across all six locales (sleepScore kept distinct from the sleep
metric's own body), wired through SubPageShell's explainerMetric prop.

The explainer-key guard resolves the score page's closed lookup record
statically so the single dynamic prop stays verifiable.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

src/components/insights/derived/composite-score-anatomy.tsx

@@ -1,8 +1,11 @@
 "use client";
 
+import { Sparkles } from "lucide-react";
+
 import { useTranslations } from "@/lib/i18n/context";
 import { useDerivedMetric } from "./use-derived-metric";
 import { Skeleton } from "@/components/ui/skeleton";
+import { InsightStatusCard } from "@/components/insights/insight-status-card";
 import type {
   ReadinessValue,
   SleepScoreValue,
@@ -193,21 +196,43 @@ export function CompositeScoreAnatomy({
     }
   }
 
+  // v1.15.12 B3 — the per-score AI assessment ("why is this score what it is")
+  // already rides the single `/api/insights/derived` route as `data.assessment`
+  // (shipped to iOS in v1.13.2); the web simply discarded it. Render it BELOW
+  // the anatomy card, reusing the same `<InsightStatusCard>` the metric
+  // sub-pages mount so it reads identically. The route always populates a
+  // non-empty assessment when the score `status === "ok"` (the deterministic
+  // template at minimum, warmer AI prose when a provider is configured), so a
+  // present `assessment` implies a real text — `hasProvider` is held true and
+  // the card always renders its prose rather than the no-provider fallback.
+  const assessment = data.assessment;
+
   return (
-    <ScoreAnatomyView
-      title={title}
-      score={score}
-      hue={METRIC_HUE[metric]}
-      caption={caption}
-      contributors={contributors}
-      coverage={data.coverage}
-      confidence={data.confidence}
-      provenance={data.provenance}
-      method={method}
-      standard={standard}
-      insufficient={insufficient}
-      className={className}
-    />
+    <div className="space-y-3">
+      <ScoreAnatomyView
+        title={title}
+        score={score}
+        hue={METRIC_HUE[metric]}
+        caption={caption}
+        contributors={contributors}
+        coverage={data.coverage}
+        confidence={data.confidence}
+        provenance={data.provenance}
+        method={method}
+        standard={standard}
+        insufficient={insufficient}
+        className={className}
+      />
+      {assessment ? (
+        <InsightStatusCard
+          title={t("insights.assessmentTitle")}
+          icon={<Sparkles className="h-5 w-5" />}
+          text={assessment.text}
+          hasProvider
+          updatedAt={assessment.updatedAt}
+        />
+      ) : null}
+    </div>
   );
 }
 

src/components/insights/derived/use-derived-metric.ts

@@ -8,6 +8,7 @@ import type {
   DerivedConfidence,
   DerivedProvenance,
 } from "@/lib/insights/derived/types";
+import type { DerivedAssessment } from "@/lib/insights/derived/derived-assessment";
 
 /**
  * v1.10.0 — the single TanStack Query hook for the generic derived-metric
@@ -44,6 +45,17 @@ export interface DerivedMetricResponse<T> {
   confidence: DerivedConfidence | null;
   provenance: DerivedProvenance;
   reason: string | null;
+  /**
+   * v1.13.2 — the per-score "why is this score what it is" assessment
+   * ({ text, source, updatedAt }). Populated only for the assessable scores
+   * (READINESS, SLEEP_SCORE, RECOVERY_SCORE, STRAIN_SCORE, STRESS_SCORE) when
+   * `status === "ok"`, null otherwise. The single-metric route serves the
+   * AI-warm prose (falling back to the deterministic template); the batch route
+   * serves the cheap deterministic text only. Optional on the wire shape so
+   * grid reads that drop it stay typed. v1.15.12 B3 surfaces it on the web
+   * score-anatomy detail page (it already shipped to iOS in v1.13.2).
+   */
+  assessment?: DerivedAssessment | null;
 }
 
 export interface UseDerivedMetricOptions {