Sparkling Water cluster name should contain spark app id instead of random number

This is to be consistent with , where logs are put into directories named by spark app id