Including SplitByRG in the FullProcessingPipeline

Why wasn't it there before, you ask
----------------------------------

Before I was running it separately (by hand), but now it's integrated in
the FullProcessingPipeline.

Integration was a pain because of Queue's limitation of only allowing 1
@Output file. This forced me to write the ugliest piece of code of my
life, but it's working and it's processing the YRI from scratch using
that right now. So I'm happy... somewhat.

Other changes to the pipeline
-----------------------------

   * Add --filter_bases_not_stored to the IndelRealigner step -- sometimes BAM files have reads with no bases stored in the unmapped section (no idea why) but this disrupts the pipeline.
   * Change adaptor marking parameter to "dual indexed" instead of "pair-ended" -- for PCR Free data.

public/scala/src/org/broadinstitute/sting/queue/util/QScriptUtils.scala

@@ -73,7 +73,7 @@ object QScriptUtils {
   /**
    * Check if there are multiple samples in a BAM file
    */
-  def hasMultipleSamples(readGroups: java.util.List[SAMReadGroupRecord]): Boolean = {
+  def hasMultipleSamples(readGroups: Seq[SAMReadGroupRecord]): Boolean = {
     var sample: String = ""
     for (r <- readGroups) {
       if (sample.isEmpty)