Quieting Spark local context logs in tests.

Merge pull request #7 from AlpineNow/jt-update-spark-csv-version

DEV-14701 Updating the Spark CSV Version to 1.3, and marking that as provided, along with the spark-avro jar.

DEV-13397 Handle cases where training/validation datasets end up being empty by throwing proper exceptions.

Merge branch 'GradientBoosting_and_Refactoring'

DEV-12802 An addition of Gradient Boosting. Additionally, a massive refactoring of the previous Random Forest code. Removed sbt in favor of pom.xml. GBT and RF share the decision tree codebase.

Changing groupby to reducebykey in the distributed node splits.

Removing some horrible inefficiencies from the equal frequency discretizer.

Improving the performance of the node Id RDD through constant re-caching.

Working checkpoint deletions.

Fixing problems with checkpoint deleting.

Adding a cache to remember the heaviest child node when predicting.

Fixing string split problems when there are trailing empty columns in rows.

Changing the default delimiter to be tab for transformer runners.

Adding some fixes for handling empty string handling.