prio: backlog: logger? slog? error handling sort by column header not only by data key? transformer in config -> manipulate source better in memory persistence? https://github.com/kelindar/column grok parser?