Re: ole32: Return the correct error in FileLockBytesImpl_*LockRegion*().

[Vincent Povirk]

It looks to me like locks that fail due to conflicts will set
ERROR_LOCK_VIOLATION, not ERROR_ACCESS_DENIED. We should probably add
some tests for the error to the LockFileEx tests.

I'd also be interested in what Windows returns on a filesystem that
doesn't support locking, if such a thing exists.