[SPARK-7384][Core][Tests] Fix flaky tests for distributed mode in BroadcastSuite

Fixed the following failure: https://amplab.cs.berkeley.edu/jenkins/job/Spark-1.3-Maven-pre-YARN/hadoop.version=1.0.4,label=centos/452/testReport/junit/org.apache.spark.broadcast/BroadcastSuite/Unpersisting_HttpBroadcast_on_executors_and_driver_in_distributed_mode/

The tests should wait until all slaves are up. Otherwise, there may be only a part of `BlockManager`s registered, and fail the tests.

Author: zsxwing <zsxwing@gmail.com>

Closes #5925 from zsxwing/SPARK-7384 and squashes the following commits:

783cb7b [zsxwing] Add comments for _jobProgressListener and remove postfixOps
1009ef1 [zsxwing] [SPARK-7384][Core][Tests] Fix flaky tests for distributed mode in BroadcastSuite