doc_cfg should not merge cfgs on non-glob inlined reexports

[Nemo157]

I tried this code:

#![feature(doc_auto_cfg)]

#[cfg(a)]
pub mod a {
    pub use crate::b::B;
}

#[cfg(any(a, b))]
mod b {
    pub struct B;
}

I expected to see this happen: The docs show that a::B requires the cfg a only.

Instead, this happened: The docs show that a::B requires the cfg a and (a or b).

Reasoning: When re-exporting things with different cfgs like this there are two things that can happen:

1. The re-export uses a subset of cfgs like the example, this subset is sufficient so that the item will appear exactly with the subset
2. The re-export uses a non-subset of cfgs (e.g. cfg(any(a, c)) on mod a above), if the non-subset cfgs are active (--cfg=c) then this will be a compile error as the item doesn't exist to re-export, if the subset cfgs are active it behaves like 1.

This only applies to non-glob inlined re-exports, for glob re-exports the item may or may not exist to be re-exported (potentially the cfgs on the path up until the glob can be removed, and only cfgs on the globbed item itself matter, but I haven't thought through all the details), for non-inlined re-exports see #85043.

[GuillaumeGomez]

Just to be sure, you mean that only the cfgs on the final re-export itself should be visible for this re-export, right?

[Nemo157]

Yes