feat: add runner failover rules and cache affinity

apps/api/internal/runner/limits_test.go

@@ -68,6 +68,35 @@ func TestBillingsSplitsCachedInputTokens(t *testing.T) {
 	}
 }
 
+func TestBillingsDefaultsCachedInputPriceToOneTenthOfInputPrice(t *testing.T) {
+	service := &Service{}
+	candidate := store.RuntimeModelCandidate{
+		ModelName: "test-model",
+		BillingConfig: map[string]any{
+			"textInputPer1k":  1.0,
+			"textOutputPer1k": 2.0,
+		},
+	}
+	response := clients.Response{
+		Usage: clients.Usage{
+			InputTokens:       1000,
+			CachedInputTokens: 400,
+			OutputTokens:      500,
+			TotalTokens:       1500,
+		},
+	}
+
+	lines := service.billings(context.Background(), nil, "chat.completions", nil, candidate, response, false)
+
+	if len(lines) != 3 {
+		t.Fatalf("expected uncached input, cached input, and output lines, got %+v", lines)
+	}
+	cached, _ := lines[1].(map[string]any)
+	if cached["resourceType"] != "text_cached_input" || cached["quantity"] != 400 || cached["amount"] != 0.04 {
+		t.Fatalf("unexpected cached input fallback billing line: %+v", cached)
+	}
+}
+
 func TestEffectiveRateLimitPolicyTreatsEmptyRuntimePolicyAsUnlimited(t *testing.T) {
 	policy := effectiveRateLimitPolicy(store.RuntimeModelCandidate{
 		PlatformRateLimitPolicy: map[string]any{"rules": []any{