adventure: repair the realm-firsts ledger the backfill got wrong

The one-shot that seeded news_realm_firsts filtered on `abandoned = 0`, and
`abandoned` does not mean anybody gave up: abandonZoneRunByID exists to retire a
run whose boss is already dead when the expedition travels onward, which is how
30 of prod's 32 boss kills are stored. So the ledger holds 6 zones where the run
history knows 9, and it dates every row to the minute the job ran rather than to
the clear.

The ledger is what claimRealmFirst tiers live dispatches against, so each missed
zone is a spurious PRIORITY "realm first" waiting to fire the next time somebody
clears it, months late.

Fix the shared query, and add a re-seed (not a re-run) under its own job name: it
writes the ledger, backdated to the real earliest clear, and emits nothing at
all. It runs regardless of the news switches, because a ledger that is right only
while emission is on mis-tiers the first dispatch after somebody flips it. Kept
in place afterwards as an ordinary bootstrap for fresh deploys.

Claude-Session: https://claude.ai/code/session_012bxpQQJDjC1mTtLN3VVtBQ

internal/plugin/bootstrap_pete_news.go

@@ -58,34 +58,104 @@ func (p *AdventurePlugin) bootstrapPeteNewsBackfill() {
 		"zone_firsts", firsts, "deaths", deaths, "achievements", achv)
 }
 
-// backfillZoneFirsts seeds news_realm_firsts from history and emits one PRIORITY
-// realm-first dispatch per zone, attributed to its earliest boss-defeating
-// clearer. SQLite returns the user_id from the same row as MIN(completed_at)
-// (bare-column min/max rule), so the (zone, first clearer, time) triple is
-// consistent. Returns the count emitted.
-func (p *AdventurePlugin) backfillZoneFirsts() int {
+// zoneFirstClear is one zone's earliest boss kill: who did it and when.
+type zoneFirstClear struct {
+	zoneID, userID, completedAt string
+}
+
+// zoneFirstClears reads the earliest boss-defeating run of every zone.
+//
+// The filter is `boss_defeated = 1` and NOTHING else, and that is the whole
+// point. `abandoned` does not mean anybody gave up — abandonZoneRunByID exists
+// to retire a run whose boss is ALREADY DEAD when the expedition travels onward
+// (dnd_zone_run.go), so in prod 30 of 32 boss kills carry abandoned = 1. An
+// `AND abandoned = 0` here drew a realm where 2 zones had ever been beaten
+// instead of 9. It is the same filter loadRealmClearStats documents; do not
+// reintroduce it.
+//
+// SQLite returns the user_id from the same row as MIN(completed_at) (bare-column
+// min/max rule), so the (zone, first clearer, time) triple is internally
+// consistent.
+func zoneFirstClears() []zoneFirstClear {
 	rows, err := db.Get().Query(
 		`SELECT zone_id, user_id, MIN(completed_at)
 		   FROM dnd_zone_run
-		  WHERE boss_defeated = 1 AND completed_at IS NOT NULL AND abandoned = 0
+		  WHERE boss_defeated = 1 AND completed_at IS NOT NULL
 		  GROUP BY zone_id`)
 	if err != nil {
 		slog.Error("backfill: zone-firsts query", "err", err)
-		return 0
+		return nil
 	}
-	type first struct {
-		zoneID, userID, completedAt string
-	}
-	var firsts []first
+	defer rows.Close()
+
+	var firsts []zoneFirstClear
 	for rows.Next() {
-		var f first
+		var f zoneFirstClear
 		if err := rows.Scan(&f.zoneID, &f.userID, &f.completedAt); err != nil {
 			slog.Error("backfill: zone-firsts scan", "err", err)
 			continue
 		}
 		firsts = append(firsts, f)
 	}
-	rows.Close()
+	return firsts
+}
+
+// bootstrapRealmFirstsReseed repairs the zone half of news_realm_firsts.
+//
+// The ledger is what claimRealmFirst tiers live dispatches against, so a zone
+// whose first clear the original one-shot missed is a spurious PRIORITY "realm
+// first" waiting to fire the next time somebody clears it, months after the
+// fact. Two things were wrong with what got seeded:
+//
+//  1. The `abandoned = 0` filter above, which is why prod holds 6 zones where
+//     the run history knows 9.
+//  2. first_at is claimRealmFirst's unixepoch() — when the claim was RECORDED,
+//     not when the clear happened. Every backfilled prod row carries the one
+//     minute the job ran.
+//
+// This is a re-SEED, not a re-run: it writes the ledger and emits nothing at
+// all, so no historical realm-first dispatch reaches the room. It has its own
+// job name because the original one-shot's gate is already marked, and per
+// feedback_loader_rewire_needs_bootstrap it stays in place afterwards — a fresh
+// deploy runs it as an ordinary bootstrap.
+//
+// It runs unconditionally on the news seam's switches, unlike the backfill: a
+// ledger that is correct only when emission happens to be on is a ledger that
+// mis-tiers the first dispatch after somebody flips it.
+//
+// A zone claim with no surviving run behind it is left exactly as it is. The run
+// history is the better record of both who and when, but only where it has one.
+func bootstrapRealmFirstsReseed() {
+	const jobName = "pete_realm_firsts_reseed_v1"
+	if db.JobCompleted(jobName, "once") {
+		return
+	}
+
+	seeded := 0
+	for _, f := range zoneFirstClears() {
+		ts, ok := parseSQLiteTime(f.completedAt)
+		if !ok {
+			slog.Warn("reseed: unparseable clear time", "zone", f.zoneID, "at", f.completedAt)
+			continue
+		}
+		// Upsert, not INSERT OR IGNORE: the six rows that already exist carry the
+		// wrong date and correcting them is half of what this job is for.
+		db.Exec("realm-firsts reseed",
+			`INSERT INTO news_realm_firsts (kind, target, first_at) VALUES ('zone', ?, ?)
+			 ON CONFLICT(kind, target) DO UPDATE SET first_at = excluded.first_at`,
+			f.zoneID, ts.Unix())
+		seeded++
+	}
+
+	db.MarkJobCompleted(jobName, "once")
+	slog.Warn("bootstrap: realm-firsts ledger reseeded", "zones", seeded)
+}
+
+// backfillZoneFirsts seeds news_realm_firsts from history and emits one
+// dispatch per zone, attributed to its earliest boss-defeating clearer.
+// Returns the count emitted.
+func (p *AdventurePlugin) backfillZoneFirsts() int {
+	firsts := zoneFirstClears()
 
 	n := 0
 	for _, f := range firsts {