Release 0.4.0

[harish]

Release 0.4.0 is now GA. 

Some of the enhancements are:

- Source Tables and Druid DataSource can be one or more databases

- Support Spark 1.6.1, 1.6.2 and Druid 0.9.1; There is now a Spark Shim Layer and we have released versions for both Spark 1.6.1 and 1.6.2

- Support for ThriftServer HA

- A SparklineData Plugin Module framework. (For e.g. see A Mysql Compatibility Sample Module)

- Support for Spatial Queries, Approximate HLL aggregators.

- A Druid TestCluster Framework; now you can download the source code and run all the tests w/o having to setup your Druid Test Cluster.

- Many more features and bug fixes, see the Release Page for details.

This will be the last Spark1.6.x release. New features will be developed on Spark 2.x