feature: improvements for infrahubctl `dump` and `load`

[ajtmccarty]

Component

Python SDK, infrahubctl CLI

Describe the Feature Request

We can improve the infrahubctl dump and load commands in number of ways

• [ ] opsmill/infrahub#2432
• [x] opsmill/infrahub#2431
• [ ] get smaller versions of objects to export instead of fully-hydrated versions. We don't always need the fully-hydrated version of an object. Being smarter about which parts of which objects we retrieve would increase performance of the commands
• [ ] support parquet. pyArrow docs on parquet (https://arrow.apache.org/cookbook/py/io.html#reading-a-parquet-file)
• [ ] better format for exporting/importing (something more like a pyarrow table for each kind of node). Current format is basically 3 columns: ID, kind, JSON dump of the GraphQL query used to create the item. If we parsed the GraphQL and saved/read it in a smarter way, we could improve performance
• [ ] save export as a compressed directory. Right now we save nodes.json to a directory when we run dump and load expects the path of the directory as a command arg. We could just compress the directory and pass that around instead of an actual directory

Describe the Use Case

No response

Additional Information

No response

[ryanmerolle]

This would be useful for sharing seed data sort of like the netbox devicetypes repo or prepopuating readable demo/seed data without all the bloat of hidden and such.