fix: SQLite connection leak causing unbounded memory growth

SaveStats returned tx.Create().Error directly without assigning to the
local err variable, so the deferred closure always saw err==nil and
called Commit() on a failed transaction. This left the underlying
go-sqlite3 connection unreturned to the pool. With StatsJob firing
every 10s, leaked connections accumulated ~150 KB each, reaching
400+ MB after ~33 hours.

Fixes:
- stats.go: assign Create result to err so defer can Rollback on failure
- backup.go: defer-close backupDb to prevent pool leak on early return
- migration/main.go: defer-close migration db
- db.go: add ConnMaxIdleTime(5m), lower MaxIdleConns to 2, set
  _cache_size=-200 to reduce per-connection memory from ~2 MB to ~200 KB

Measured: RSS dropped from 419 MB to 66 MB, db file descriptors from
4484 to 6, with zero growth over 3-minute observation window.

database/db.go

@@ -55,7 +55,10 @@ func OpenDB(dbPath string) error {
 	if strings.Contains(dbPath, "?") {
 		sep = "&"
 	}
-	dsn := dbPath + sep + "_busy_timeout=10000&_journal_mode=WAL"
+	// _cache_size=-200 caps each connection's page cache at ~200 KiB
+	// (default is ~2 MiB), reducing memory amplification if a connection
+	// escapes the pool.
+	dsn := dbPath + sep + "_busy_timeout=10000&_journal_mode=WAL&_cache_size=-200"
 	db, err = gorm.Open(sqlite.Open(dsn), c)
 	if err != nil {
 		return err
@@ -66,8 +69,9 @@ func OpenDB(dbPath string) error {
 		return err
 	}
 	sqlDB.SetMaxOpenConns(25)
-	sqlDB.SetMaxIdleConns(5)
+	sqlDB.SetMaxIdleConns(2)
 	sqlDB.SetConnMaxLifetime(time.Hour)
+	sqlDB.SetConnMaxIdleTime(5 * time.Minute)
 
 	if config.IsDebug() {
 		db = db.Debug()