Support Dot Product for Vectors

I believe org.apache.spark.ml.linalg.Vectors and org.apache.spark.mllib.linalg.Vectors should support the dot product. The necessary BLAS routines are already there, just a simple wrapper is needed.

I know there has been a lot of discussion about how much of a linear algebra package should spark attempt to be, but I have found that the dot product comes up quite a bit for feature engineering and scoring. In the past we have created our own org.apache.spark.ml.linalg package to expose the private methods, but it's an annoying hack.

See also:

SPARK-6442
SPARK-10989