[feat]: ultimate normalization script

Going forward, there are likely many times we may want to normalize records. As such, we need a smart script:

**Musts**

- [ ] not bring down production to run script
- it should be a script we can tune to **not** hammer the db, and quietly run in the background
- [ ] easy to define/add transformations
- [ ] fault tolerant
- [ ] keep track of versioning
- must not repeat work - the script should be stopped/started without hesitation, and start off where it left off
- [ ] fearlessly concurrent
- the api will write a `lastUpdated` field to the user record which should allow all ops to be done without losing data

**Niceties**

- [ ] dynamically asks database how loaded it is, and accordingly adjusts ops/rps

**Notes**

Currently, the idea is to add a `schemaVersion` field to the `user` record.

I would prefer to write this in Rust:
1) Ease of distribution
2) Strict typing

That said, this might go against _"easy to define/add transformations"_.

No files, tests, or entry points are named. Start by clarifying the database and API interfaces, the proposed user.schemaVersion behavior, and how transformations are defined and versioned. Done should include resumable, fault-tolerant, concurrency-safe execution that can be throttled without repeated work or production impact.

Written by the indexing model from the issue text.