Ensure best practices are in use for generation of unique identifiers

We're currently generating identifiers using rng distributions from the `rand` crate. I'd like to assess whether or not this is sufficient for our needs. The likelihood of collision is incredibly low, but I'd like to ensure we're doing all we can while still performing well.

[ThreadRng](https://rust-random.github.io/rand/rand/rngs/struct.ThreadRng.html)

If not, we could use something like [ulid](https://crates.io/crates/ulid).

No source file, test, or entry point is named. Start by locating identifier generation that uses the rand crate and review the linked ThreadRng documentation, then compare it with ulid. Done means documenting whether the current approach meets the project's collision and performance needs and selecting an appropriate approach if it does not.

Written by the indexing model from the issue text.