Docs for task::spawn must mention support for tasks that block

[jimblandy]

The async-std blog post "stop worrying about async blocking" explains that spawn_blocking is no longer needed, and that developers should simply use spawn instead and let the library decide whether it should keep the thread in the async pool.

If task::spawn is the recommended way to spawn tasks that may block, then users will write code that depends on that, and the design decision becomes something async-std can't revert without introducing bugs in its users. This is all fine (it seems like a good design to me), but non-obvious characteristics covered by async-std's stability guarantees should be documented.

[Keruspe]

Actually, this behaviour was never actually merged #631

What I'm working on is an opt-in integration, disabled by default.

[jimblandy]

Since this isn't going to be the default behavior, this PR's documentation isn't appropriate. Documentation should be addressed as an integral part of whatever PR does get merged.