rustdoc: `#[doc(notable_trait)]` doesn't work on re-exports

[camelid]

This code correctly triggers the "Notable traits" dialog:

#![feature(doc_spotlight)]

mod my_module {
    #[doc(spotlight)]
    pub trait MyTrait {}
}

pub use my_module::MyTrait;

pub struct MyStruct;
impl MyTrait for MyStruct {}

pub fn my_function() -> MyStruct { MyStruct }

But this does not:

#![feature(doc_spotlight)]

mod my_module {
    pub trait MyTrait {}
}

#[doc(spotlight)]
pub use my_module::MyTrait;

pub struct MyStruct;
impl MyTrait for MyStruct {}

pub fn my_function() -> MyStruct { MyStruct }

The difference is that the first one puts #[doc(spotlight)] on the trait,
while the second puts it on the re-export. I'm not certain that this is supposed
to work, but there is code in src/librustdoc/clean/inline.rs, which seems to
be used for re-exports, that supposedly detects #[doc(spotlight)]:

rust/src/librustdoc/clean/inline.rs

Lines 194 to 212 in 4253153

	crate fn build_external_trait(cx: &DocContext<'_>, did: DefId) -> clean::Trait {
	let trait_items =
	cx.tcx.associated_items(did).in_definition_order().map(|item| item.clean(cx)).collect();
	let predicates = cx.tcx.predicates_of(did);
	let generics = (cx.tcx.generics_of(did), predicates).clean(cx);
	let generics = filter_non_trait_generics(did, generics);
	let (generics, supertrait_bounds) = separate_supertrait_bounds(generics);
	let is_spotlight = load_attrs(cx, did).clean(cx).has_doc_flag(sym::spotlight);
	let is_auto = cx.tcx.trait_is_auto(did);
	clean::Trait {
	unsafety: cx.tcx.trait_def(did).unsafety,
	generics,
	items: trait_items,
	bounds: supertrait_bounds,
	is_spotlight,
	is_auto,
	}
	}

That build_external_trait function is called in two places, one of which is in
the try_inline function:

rust/src/librustdoc/clean/inline.rs

Lines 58 to 62 in 4253153

	Res::Def(DefKind::Trait, did) => {
	record_extern_fqn(cx, did, clean::TypeKind::Trait);
	ret.extend(build_impls(cx, Some(parent_module), did, attrs));
	clean::TraitItem(build_external_trait(cx, did))
	}

This was discovered in #80965 (comment).

cc @jyn514

[booleancoercion]

I think you may have pasted the same code snippet twice, I can't find a difference...

[camelid]

Oops, fixed! Thanks :)