feat(api): read-all takes an optional up_to cutoff (M489)

POST /api/me/notifications/read-all accepts {"up_to": RFC3339}. Android
queues "mark all read" for replay when offline, and a replay landing later
must not mark notices that arrived in between, which the user never saw.
A coalesced notice updated since then has a newer created_at, so it stays
unread. An empty body still marks everything.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

internal/db/dbq/notifications.sql.go

@@ -214,11 +214,22 @@ func (q *Queries) ListNotifications(ctx context.Context, arg ListNotificationsPa
 const markAllNotificationsRead = `-- name: MarkAllNotificationsRead :execrows
 UPDATE user_notifications
    SET read_at = now()
- WHERE user_id = $1 AND read_at IS NULL
+ WHERE user_id = $1
+   AND read_at IS NULL
+   AND ($2::timestamptz IS NULL OR created_at <= $2)
 `
 
-func (q *Queries) MarkAllNotificationsRead(ctx context.Context, userID pgtype.UUID) (int64, error) {
-	result, err := q.db.Exec(ctx, markAllNotificationsRead, userID)
+type MarkAllNotificationsReadParams struct {
+	UserID pgtype.UUID
+	UpTo   pgtype.Timestamptz
+}
+
+// up_to, when set, limits it to what existed when the user asked: a "mark
+// all read" queued offline and replayed later must not mark notices that
+// arrived in between, which the user never saw. A coalesced row updated
+// since then carries a newer created_at, so it stays unread too.
+func (q *Queries) MarkAllNotificationsRead(ctx context.Context, arg MarkAllNotificationsReadParams) (int64, error) {
+	result, err := q.db.Exec(ctx, markAllNotificationsRead, arg.UserID, arg.UpTo)
 	if err != nil {
 		return 0, err
 	}