Bug fix: SPARK-837

Build changes for ML lib

Added property 'spark.executor.uri' for launching on Mesos.

Use the Char version of split() instead of the String one in MLUtils

Compression codec change

Refactored Vector.apply(length, initializer) replacing excessive code with library method

(also removed unused variable ```ans``` as minor change)

Cleaned Partitioner & PythonPartitioner source by taking out non-related logic to Utils

Add Apache license header to metrics system

Also don't build an assembly jar for them

Renamed compressionOutputStream and compressionInputStream to compressedOutputStream and compressedInputStream.

Conflicts:
	project/SparkBuild.scala

Use Bootstrap progress bars in web UI

Minor mllib cleanup

Also add jblas dependency to mllib pom.xml

Renamed Classification.scala to ClassificationModel.scala and Regression.scala to RegressionModel.scala

made SimpleUpdater consistent with other updaters

requiring Spark to be installed. Using 'make_distribution.sh' a user
can put a Spark distribution at a URI supported by Mesos (e.g.,
'hdfs://...') and then set that when launching their job. Also added
SPARK_EXECUTOR_URI for the REPL.