Regression: default webui-port can't be set via command line "--webui-port" anymore

Move ML lib data generator files to util/

Do not copy local jars given to SparkContext in yarn mode

Do not copy local jars given to SparkContext in yarn mode since the Context is not running on local. This bug causes failure when jars can not be found. Example codes (such as spark.examples.SparkPi) can not work without this fix under yarn mode.

Updates to LogisticRegression

[BUGFIX]  Fix for sbt/sbt script SPARK_HOME setting

fix a bug in build process that pulls in two versions of ASM.

Consistently invoke bash with /usr/bin/env bash in scripts to make code ...

Consistently invoke bash with /usr/bin/env bash in scripts to make code more portable (JIRA Ticket SPARK-817)

Also ensure weights are initialized to a column vector.

multiple train methods.

Also move LogisticGradient to the LogisticRegression file and fix the
unit tests log path.

In some environments, this command

    export SPARK_HOME=$(cd "$(dirname $0)/.."; pwd)

echoes two paths, one by the "cd ..", and one by the "pwd". Note the resulting
erroneous -jar paths below:

    ctn@ubuntu:~/src/spark$ sbt/sbt
    + EXTRA_ARGS=
    + '[' '' '!=' '' ']'
    +++ dirname sbt/sbt
    ++ cd sbt/..
    ++ pwd
    + export 'SPARK_HOME=/home/ctn/src/spark
    /home/ctn/src/spark'
    + SPARK_HOME='/home/ctn/src/spark
    /home/ctn/src/spark'
    + export SPARK_TESTING=1
    + SPARK_TESTING=1
    + java -Xmx1200m -XX:MaxPermSize=350m -XX:ReservedCodeCacheSize=128m -jar /home/ctn/src/spark /home/ctn/src/spark/sbt/sbt-launch-0.11.3-2.jar
    Error: Invalid or corrupt jarfile /home/ctn/src/spark

Committer: ctn <ctn@adatao.com>

On branch master
Changes to be committed:

- Send output of the "cd .." part to /dev/null
	modified:   sbt/sbt

Dependency upgrade Akka 2.0.3 -> 2.0.5

Conflicts:
	make-distribution.sh

Throw a more meaningful message when runJob is called to launch tasks on non-existent partitions.

Link to job UI from standalone deploy cluster web UI