make accumulator.localValue public, add tests

Changes to SizeEstimator more accurate

oops setting to get deterministic results

setting. This is useful to get a deterministic test case

according to architecture and also aligns objects and arrays when computing
instance sizes. Verified using Eclipse Memory Analysis Tool (MAT)

Examples ship to to cluster

This is because we pull out such a large sample (10x the number of
partitions) that we should expect pretty good balance. The tests are
also deterministic so there's no worry about them failing irreproducibly.

RangePartitioner to make it pass. It turns out that the first partition
was always kind of small due to how we picked partition boundaries.

Bug fix in RangePartitioner for partitioning when sorting in descending order.

Logging Throwables in Info and Debug

Logging Throwables in logInfo and logDebug instead of swallowing them.

add Accumulatable, add corresponding docs & tests for accumulators

Always destroy SparkContext in after block for the unit tests.

Instantiating custom serializer using user's classpath