Concurrency fix for MacOS

[sagacity]

This is a fix for #11.

I needed to wrap all NSPasteboard access in this mutex, including the new() fn. I've used a global mutex for this, wrapped in a lazy_static!.

Edit: Hmm, no, apparently even WITH a Mutex around it, it's still failing intermittently :( It appears that even a global Mutex isn't enough, it's simply failing even when multiple pasteboard instances are being created and used. The only way I can get my test to reliably work is if have a shared Arc<Mutex<ClipboardContext>> and re-use that everywhere. Simply creating multiple ClipboardContexts will eventually fail.

[chrisduerr]

Hmm, no, apparently even WITH a Mutex around it, it's still failing intermittently :( It appears that even a global Mutex isn't enough, it's simply failing even when multiple pasteboard instances are being created and used.

Have you checked Apple's documentation w.r.t. NSPasteboard? I feel like this is an issue that should be documented somewhere and gaining a better understanding of what is actually going on would probably help fix this.

The only way I can get my test to reliably work is if have a shared Arc<Mutex> and re-use that everywhere.

Well in theory if this really is our only option, it should be possible to just always give out a reference to a singleton which is managed with a lock in the background.

[sagacity]

Let's first do a root cause analysis...