Investigate if this can happen in production: someone does a backup to some data and then tries to do a PIT restore to immediately after, so that the config server (or any server) has no new writes in a sharded cluster.
If this is possible, it would hit fassert(8290701, reader.hasNext()).