configuring local block size for spark if not using hadoop

Following #269, what if user isn't using hadoop and only using spark & spark scheduler?

I bet that if you create the hadoop core files (e.g. core-site.xml) and put it in Spark's path, it'll work.  But there's no mechanism for that now if hadoop isn't enabled.  Some re-architecture would be needed.  This may be a Magpie 2.0 issue.