Prefer doc(no_inline) to doc(hidden) for reexports

[nipunn1313]

Fixes #2472

[nipunn1313]

The cargo clippy failure appears to be a regression introduced in nightly 2021-08-17. Afaict it's a false positive from clippy.

The failure of cargo +1.41.0 is interesting since it's clearly ok on some stables around it. It's also clearly a non-regression (claiming that things aren't documented there). What do you suggest we do here to make CI happy? Could conditionally ignore the warning with something like https://github.com/dtolnay/rustversion - or unconditionally ignore the warning.

[taiki-e]

Thanks! Could you change to use no_inline only on trait reexports? *_internal macros and async_await module are not public APIs.

[taiki-e]

Could you do the same for the trait reexports in futures?

futures-rs/futures/src/lib.rs

Lines 105 to 125 in fdb96f5

	#[doc(hidden)]
	pub use futures_core::future::{Future, TryFuture};
	#[doc(hidden)]
	pub use futures_util::future::{FutureExt, TryFutureExt};
	#[doc(hidden)]
	pub use futures_core::stream::{Stream, TryStream};
	#[doc(hidden)]
	pub use futures_util::stream::{StreamExt, TryStreamExt};
	#[doc(hidden)]
	pub use futures_sink::Sink;
	#[doc(hidden)]
	pub use futures_util::sink::SinkExt;
	#[cfg(feature = "std")]
	#[doc(hidden)]
	pub use futures_io::{AsyncBufRead, AsyncRead, AsyncSeek, AsyncWrite};
	#[cfg(feature = "std")]
	#[doc(hidden)]
	pub use futures_util::{AsyncBufReadExt, AsyncReadExt, AsyncSeekExt, AsyncWriteExt};

[nipunn1313]

yep - thanks for catching! I missed that file in my audit! (I had used vim -p $(rg "doc.*hidden") which capped to an upper bound number of files by mistake). I double checked and confirmed I covered everything after hitting up futures/src/lib.rs

Thanks!

[taiki-e]

Thanks!