Bulk CSV Load doesn't work as expected

Going by this documentation
https://age.apache.org/age-manual/master/intro/agload.html#explanation-about-the-csv-format.

I started playing around a little over a week ago with AGE - programmatically populated over 1.5M nodes so far and (10x + on relations). Figured that's not sustainable and wanted to pivot to bulk load.

I have a vertex file with a handful of nodes and random ids (deliberately didn't start with 1, but they are definitely matching the edges). the nodes are successfully imported. however, the edges don't work out at all. based on documentation, the 'custom_id' need to match between nodes and edges, which it does, but upon importing the edges, the start_id and end_id point to random internal ids of nodes which I have no clue how to resolve.

[HAS_PROGRAM.csv](https://github.com/user-attachments/files/19471808/HAS_PROGRAM.csv)
[HAS_TAG.csv](https://github.com/user-attachments/files/19471809/HAS_TAG.csv)
[nodes.csv](https://github.com/user-attachments/files/19471810/nodes.csv)