Also add multiple versions of them to make it easier to call them from java.

Random seeds make more sense by default for a machine learning library
because other libraries behave the same way (people expect to be able to
run the algorithm multiple times and get a better answer), but we can
add configuration later if needed. Tests that depend on specific seed
choices seem brittle.

Before, when withReplacement was set to true, we would not get a sample
bigger than the RDD's count().

Conflicts:
	core/src/main/scala/spark/RDD.scala
	core/src/test/scala/spark/RDDSuite.scala

Includes unit tests and sample data to run on.

Also add a feature to get the intercept correct and test these
using a small unit test.

Remove extraneous methods and classes

Verified that results match with glm in R

Also refactor RidgeRegression and LogisticRegression to re-use code
and update the test as well

Data from
http://www-stat.stanford.edu/~tibs/ElemStatLearn/datasets/prostate.data

Also update model after training based appropriately.

this :s.

Conflicts:
	run

Clean StageToInfos periodically when spark.cleaner.ttl is enabled

hbase dependency is missed in hadoop2-yarn profile of examples module