[ASDisplayNode] Convert isSynchronous to an Objective-C atomic BOOL.

[appleguy]

This reduces lock contention, and should also fix a very rarely seen deadlock.

See #64 for details.

@Adlai-Holler -- this is a case where I think a std::atomic would probably be better if we had a convenient access pattern. If we carried this pattern to other flags, the message send overhead would definitely become significant.

Although it's in the margin (most apps spend between 8-13% of all CPU time in objc_msgSend), for a core library it's great to keep that overhead as low as practical; BOOL flags seem like a possible opportunity to do something really efficient.

I'm also wondering if we could use a bitfield or something else more packed. It's not an issue for one BOOL, but I bet we'd get significant gains out of moving all the _flags to std::atomic bitfields if they were kept within 32 bits per field. We'd be able to avoid locking the mutex in many situations we do today, which would help not just CPU time but also contention / scheduler waiting.

[ghost]

	1 Warning
⚠️	Any source code changes should have an entry in CHANGELOG.md or have #trivial in their title.

Generated by 🚫 Danger

[Adlai-Holler]

Great! Let's standardize on using property names, instead of accessor names, with dot syntax, so self.isSynchronous -> self.synchronous OR [self isSynchronous].

[appleguy]

Here's a very rough cut at what a better _atomicFlags setup might look like:

https://gist.github.com/appleguy/9b7c325b1678d27312f33abaea21d2ce

[appleguy]

Closing in favor of #89