For most use cases of `flock`, it's very important that the lock file _not_ be "cleaned up". Otherwise, imagine this scenario:
* process A opens the lock file, finds it does not exists, so it creates it.
* process A acquires the lock
* process B opens the lock (finds it already exists)
* process B tries to acquire the lock but has to wait
* process A releases the lock
* process B acquires the lock instantly
* process A deletes the lock file
* process C opens the lock file, finds it does not exists, so it creates a new one. Note that it is now holding open a _different_ lock file that the one that process B has locked.
* process C tries to acquire the lock and succeeds... but it should have had to wait, because process B still has [a prior incarnation of] the lock file open and locked.