[MERGE] Add an optional 'token' keyword argument to LockableFiles.lock_write

[Martin Pool]

Andrew Bennetts wrote:
> Collisions aren't a concern; the nonce is 20 bytes of randomness already, and I
> believe that's all the existing lock code compares anyway.

ok

> The debugging point is a good one though.  I need to see what's needed on the
> smart server branch to make sure useful information gets put in the info, and
> then returned to the client in the case of contention.

OK.  For real luxury the lock info could contain information about both
the client and server processes.  At any rate the client should pass
this when taking the lock.

-- 
Martin