[SPARK-6141][MLlib] Upgrade Breeze from 0.10 to 0.11 to fix convergence bug

LBFGS and OWLQN in Breeze 0.10 has convergence check bug.
This is fixed in 0.11, see the description in Breeze project for detail:

scalanlp/breeze#373 (comment)

Author: Xiangrui Meng <[email protected]>
Author: DB Tsai <[email protected]>
Author: DB Tsai <[email protected]>

Closes #4879 from dbtsai/breeze and squashes the following commits:

d848f65 [DB Tsai] Merge pull request #1 from mengxr/AlpineNow-breeze
c2ca6ac [Xiangrui Meng] upgrade to breeze-0.11.1
35c2f26 [Xiangrui Meng] fix LRSuite
397a208 [DB Tsai] upgrade breeze

(cherry picked from commit 76e20a0)
Signed-off-by: Xiangrui Meng <[email protected]>

Author: mengxr

mllib/pom.xml

@@ -63,7 +63,7 @@
     <dependency>
       <groupId>org.scalanlp</groupId>
       <artifactId>breeze_${scala.binary.version}</artifactId>
-      <version>0.10</version>
+      <version>0.11.1</version>
       <exclusions>
         <!-- This is included as a compile-scoped dependency by jtransforms, which is
              a dependency of breeze. -->

mllib/src/test/scala/org/apache/spark/mllib/classification/LogisticRegressionSuite.scala

@@ -372,8 +372,12 @@ class LogisticRegressionSuite extends FunSuite with MLlibTestSparkContext with M
     testRDD2.cache()
     testRDD3.cache()
 
+    val numIteration = 10
+
     val lrA = new LogisticRegressionWithLBFGS().setIntercept(true)
+    lrA.optimizer.setNumIterations(numIteration)
     val lrB = new LogisticRegressionWithLBFGS().setIntercept(true).setFeatureScaling(false)
+    lrB.optimizer.setNumIterations(numIteration)
 
     val modelA1 = lrA.run(testRDD1, initialWeights)
     val modelA2 = lrA.run(testRDD2, initialWeights)