API: Include the lint crate in the lint identifier

[xFrednet]

This PR adds the lint crate name, as an infix for the lint identifier.

• marker::diag_msg_uppercase_start -> marker::marker_lints::diag_msg_uppercase_start

This prevents lint name collisions across different lint crates.

---

This is a major breaking change, but I think it'll definitely be worth it in the long run. We might want to add a way to rename lint crates, this would allow users to shorten some names, like this:

• marker::super_lints::diag_msg_uppercase_start -> marker::super::diag_msg_uppercase_start

---

Inspired by #278

[Veetaha]

Beware that I changed the book chapter about specifying lints in #283. The docs there already include the requirement of specifying the lint crate name in the lint name.

[Veetaha]

This looks like a painful papercut. I don't fully understand how it works. Why is this marker_lints_uitests and not just marker_lints? The name marker_lints_uitests isn't mentioned anywhere else in this PR. I also don't see this code setup for marker_uilints.

Could this be avoided? Otherwise, could the workaround be encapsulated inside of marker_uitest::simple_ui_test_config?

[xFrednet]

ui_test doesn't use Cargo and instead calls the rustc driver directly. This is for a few reasons, as Cargo would need a workspace, Cargo.toml file and would also be a bit slower for only compiling a single file. However, this also means that any environment values usually set by Cargo (starting with CARGO_*) are not set when the uitest files are compiled.

This should really only be noticed, if someone writes lints for the marker_api::declare_lint! macro. That's why only the uitest of marker_lints required this environment value to be set. So I don't really see it as a paper cut.

[xFrednet]

I assume this PR is ready, since all comments have been addressed. Merging this will also unblock the docs from #283

Thank you for the review and documentation update in #283