I have a webapp on a machine implemented as a persistent process which pools open sqlite database connections without closing them on a WAL-mode database. The machine's file system is backup daily using file system backup software (namely [Borg](https://www.borgbackup.org/)). Can I assume the backup software will always snapshot a consistent database state or do I need to use one of the various available means (sqlite3 tool, vacuum, backup API) to periodically make a separate, stable, database file that can be picked up by the backup software ? (I'd rather not do that for space reasons)