Fixed ASTERIXDB-1249 and ASTERIXDB-1250: self index-nested-loop join correctly identifies the outer and the inner branch. The first dataset becomes the outer branch. The second dataset becomes the inner branch. The optimizer for index-nested-loop join now only try to use an index from the inner branch.
Run ExecutionTest w/integration test framework I changed the default local.xml to use 2 NCs with 2 partitions each like AsterixHyracksIntegrationUtil so that the results will match despite using simple string compare to determine correctness.
There is also one test in particular (big-object-join) that I had to add an artificial order by clause to to deal with different results.
Issues outstanding: - HDFS in ClusterExecutionIT needs to be thought out