Persist user units from unit archive

[Tom2096]

Summary

• add a curated unit archive loaded at server startup from archive_root/unit.json
• add UserUnitService helpers for inserting and reading DB-backed UserUnitInfo fields
• seed the first placeholder unit from the archive when a user has no local units
• enable the user_units migration and generate archive structs from packet-generator

Packet-generator schema PR: decompfrontier/packet-generator#26

Testing

• cmake --preset debug-win64
• cmake --build . --config Debug

[Tom2096]

TLDR:

• Added a /deploy/archive/unit.json that acts as our unit archive (similar to unit MST, except it has only fields that we understand/uses - extend in the future)
• Added a UserArchiver global singleton that reads unit.json and supports simple lookup APIs.
• Added a UserUnitService static class that abstracts away adding/reading units to local SQL
• Enabled the user units migration

[Tom2096]

Working - stats pulled from unit.json

[Tom2096]

Some thoughts on why I think this is the simplest approach:

• Schema consistency matters; it is the reason we use the packet generator. Reusing generated packet structs prevents separate database models and network packet models from drifting apart as the codebase evolves.

• Most database operations are triggered by fields received from the client or used to build a response. Mapping packets directly to database columns lets us read and write those fields without repeatedly implementing fragile conversion logic.

• Adding new fields to existing databases/new databases is painless: we just define the packet that interacts with the new fields, add the strict compile-time mapping, and the API will handle the rest.