The build system uses a class called Singleton to handle file locking. This class has a few potential issues:
1) An instance of Single could delete a file owned by another instance
2) Calling `print` in `__enter__` doesn't write to STDOUT
3) If the lock file already exists but the process that "owned" it died without deleting it every process will be locked
This change is an attempt to resolve these issues and provide some lock debugging code.
The code is structured to default to using the current Singleton class unless the USE_FLOCK_CLS environmental variable is set. In the event the environment variable is set the Singleton class is replaced with FLock.
The FLock class is designed to work while running on a machine that's also using the real Singleton class. It does this at the expense of not solving issue number #1. If FLock is proven successful and completely deployed to the cluster it should be made the default and the code that deletes the lock file in `__exit__` should be removed.
