[SPARK-19679][ML] Destroy broadcasted object without blocking

## What changes were proposed in this pull request?
Destroy broadcasted object without blocking
use `find mllib -name '*.scala' | xargs -i bash -c 'egrep "destroy" -n {} && echo {}'`

## How was this patch tested?
existing tests

Author: Zheng RuiFeng <ruifengz@foxmail.com>

Closes #17016 from zhengruifeng/destroy_without_block.

mllib/src/main/scala/org/apache/spark/ml/tree/impl/GradientBoostedTrees.scala

@@ -226,7 +226,7 @@ private[spark] object GradientBoostedTrees extends Logging {
       (a, b) => treesIndices.map(idx => a(idx) + b(idx)))
     .map(_ / dataCount)
 
-    broadcastTrees.destroy()
+    broadcastTrees.destroy(blocking = false)
     evaluation.toArray
   }
 

mllib/src/main/scala/org/apache/spark/mllib/optimization/LBFGS.scala

@@ -261,7 +261,7 @@ object LBFGS extends Logging {
       val (gradientSum, lossSum) = data.treeAggregate((zeroSparseVector, 0.0))(seqOp, combOp)
 
       // broadcasted model is not needed anymore
-      bcW.destroy()
+      bcW.destroy(blocking = false)
 
       /**
        * regVal is sum of weight squares if it's L2 updater;

mllib/src/main/scala/org/apache/spark/mllib/tree/model/treeEnsembleModels.scala

@@ -167,7 +167,7 @@ class GradientBoostedTreesModel @Since("1.2.0") (
       (a, b) => treesIndices.map(idx => a(idx) + b(idx)))
     .map(_ / dataCount)
 
-    broadcastTrees.destroy()
+    broadcastTrees.destroy(blocking = false)
     evaluation.toArray
   }