chore: Improve how the other suggestions message gets rendered

[Muscraft]

Note: This change is part of my ongoing work to use annotate-snippets as rustc's emitter

This change started as a way to remove some specialty code paths from annotate-snippets, by making the "and {} other candidates" message get rendered like a secondary message with no level, but turned into a fix for the message's Unicode output. Before this change, when using the Unicode output, the other suggestions message would get rendered outside of the main suggestion block, making it feel disconnected from what it was referring to. This change makes it so that the message is on the last line of the block, aligning its rendering with other secondary messages, and making it clear what the message is referring to.

Before:

error[E0433]: failed to resolve: use of undeclared type `IntoIter`
   ╭▸ $DIR/issue-82956.rs:28:24
   │
LL │         let mut iter = IntoIter::new(self);
   │                        ━━━━━━━━ use of undeclared type `IntoIter`
   ╰╴
help: consider importing one of these structs
   ╭╴
LL + use std::array::IntoIter;
   ├╴
LL + use std::collections::binary_heap::IntoIter;
   ├╴
LL + use std::collections::btree_map::IntoIter;
   ├╴
LL + use std::collections::btree_set::IntoIter;
   ╰╴
     and 9 other candidates

After:

error[E0433]: failed to resolve: use of undeclared type `IntoIter`
   ╭▸ $DIR/issue-82956.rs:28:24
   │
LL │         let mut iter = IntoIter::new(self);
   │                        ━━━━━━━━ use of undeclared type `IntoIter`
   ╰╴
help: consider importing one of these structs
   ╭╴
LL + use std::array::IntoIter;
   ├╴
LL + use std::collections::binary_heap::IntoIter;
   ├╴
LL + use std::collections::btree_map::IntoIter;
   ├╴
LL + use std::collections::btree_set::IntoIter;
   │
   ╰ and 9 other candidates

[rustbot]

r? @compiler-errors

rustbot has assigned @compiler-errors.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[jieyouxu]

Discussion: maybe use e.g. a copy of this test for diagnostics emitter testing? This const-generics test AFAICT is a regression test (for a previous ICE), I feel like it probably shouldn't be dual-serving as diagnostics emitter testing, as there are two different test intentions? 🤔

[Muscraft]

That is a very good point. I had not considered why a test exists when choosing one. I agree that it shouldn't serve the role of a regression test and a diagnostic emitter test. Looking at the other available tests, tests/ui/suggestions/too-many-field-suggestions.stderr looks to be the best candidate, but I would be happy to create a dedicated test if that is better.

[jieyouxu]

I think a dedicated test for this specific purpose is probably best, even if superficially it looks like an exact copy of another test, because even that test looks like it has a different test intention (of suggestions for field suggestions). But a test comment would clarify that so.

In that, for the intentions of this PR, the actual "error" doesn't matter as much as the suggestion format itself :)

[estebank]

Interesting. This seems to be a mild render bug for the unicode output.

[estebank]

For reference, this should have been

error[E0423]: expected function, tuple struct or tuple variant, found struct `std::collections::HashMap`
   ╭▸ $DIR/multi-suggestion.rs:17:13
   │
LL │     let _ = std::collections::HashMap();
   ╰╴            ━━━━━━━━━━━━━━━━━━━━━━━━━━━
   ╭▸ $SRC_DIR/std/src/collections/hash/map.rs:LL:COL
   │
   ╰ note: `std::collections::HashMap` defined here
help: you might have meant to use an associated function to build this type

Pre-existing. The logic for "couldn't find span" predates the unicode output.

[Muscraft]

You are correct that the logic for "couldn't find span" is the cause of the odd output here. I have a commit locally that changes how it is rendered to resemble what you expected it to be.

[estebank]

r? @estebank
@bors r+

[bors]

📌 Commit d67bf60 has been approved by estebank

It is now in the queue for this repository.