chore: resolve macros/derived-traits from crates w/ scopes rather than globally

[donaldguy]

i.e. remove depreciated use of #[macro_import] extern crate

to make this more sane for reviewing, I did commits by crate. (and can confirm that each one builds cleanly)

I'm pretty sure the contracts, eyre, indoc, and strum ones are worth taking for overall clarity-of-provenance.

---

I have more mixed feelings on the utility of scoping pretty_assertions and insta assertions.

I, in fact, tried (at least) doing:

pub use pretty_assertions::{assert_eq, assert_ne, assert_str_eq};

in src/test.rs and then it was gonna e.g. use crate::test::* or similar (crate::test::assertions::*?), but that made stuff angry about possible ambiguity (based on implications/expansion of #[test]?)

The "correct" way to do this, I think, then is to rather overrider/replace the #[test] attribute itself with e.g. a #[mise_test] (for the insta ones maybe a #[snapshot_test]) or similar

---

I was working on a larger refactor

(trying to disentangle the cli— in particular ForgeArg, ToolArg, ToolVersionType (to limit down cli-layer usage to the FromStr-ing?); ui (SingleReport and TreeItem traits) and most of the env crates from the rest, with a view towards maybe doing a separate lib.rs and minimal main.rs)

but I wore myself out and stepped back to start with offering this [and now probably lose interest for at least several days]

¯\(ツ)/¯