Add weak_count and strong_count to Rc and Arc

[scialex]

These functions allow you to see how many weak and strong references
there are to an Arc, Rc, or an rc::Weak. Due to the design of
Arc it is not possible to get the number of weak references of an
arbitrary arc::Weak. Look in arc.rs for a more in-depth explanation.

On arc::Arc and arc::Weak these operations are wait-free and atomic.

This sort of information is useful for creating dynamically cleared caches for use in OS development, for example holding pages of files in memory until the address space is needed for something else.

[emberian]

I believe the methods should explicitly be #[experimental] for now. But, this looks good to me. I'll defer final r+ to @alexcrichton or @brson, who understand the concurrency aspects far better than I.

[alexcrichton]

We've been hesitant in the past to add new inherent methods to Rc and Arc (smart pointers in general) due to how they interact with Deref and friends. For example, we have get_mut and is_unique as standalone free functions instead of inherent methods.

Could you move these methods to free functions instead?

[scialex]

@alexcrichton I could. The fact that they are for both strong and weak references would open name conflicts (Although I could just remove the Weak versions since they will likely be less useful).

The readability of these as standalone would go way down also (and I honestly doubt many people are using names like this in any case).

[alexcrichton]

Yeah it's probably fine to remove the weak versions because a failed upgrade means there are 0 strong references and an otherwise successful upgrade allows you to get the active number of strong/weak references.

I do agree that they're not quite as nice to call (as they're not methods and require an import). We haven't closely considered methods such as this in terms of stabilization just yet more than "they should be experimental free functions for now". We may investigate an extension trait, for example, later in the future, but for now free functions follow the existing design of these modules.

[scialex]

@alexcrichton updated it so that there are only functions for Rc and Arc and the functions are top level.

[tomjakubowski]

cc @steveklabnik, this PR introduces a new method on Rc for #19112 (also, "this value" 😄)