IlluminaDataProvider now supports (and prefers) BCLs and other binary data types over QSeqs.  If BCLs,Pos,Locs,Clocs, or filter files are present then IlluminaDataProvider will use these file formats.  However, if one type of data requested requires the use of QSeqs then QSeqs will be used for all data provided (to avoid unecessarily parsing extra files).  This change includes other misc. refactorings within the Illumina package.

Minor change to allow direct setting of the bait set name via a command line parameter instead of inferring it from the filename. Inference is still used if explicit name isn't passed.

Avoid String.intern in set(Mate)ReferenceName if the string can be obtained from the sequence dictionary, because intern is surprisingly slow.

Ensure position data type is in set of data types in ctors rather than in makeDataProvider(), in order to avoid ConcurrentModificationException.

*  Fixing bug on non-indexed run
*  adding utility method to copy directory tree

Eliminate SamFileHeaderMerger.addIterator, and instead add another ctor that allows for passing iterators in the ctor.
Use LinkedHashMap instead of ArrayList in SamFileHeaderMerger for speed.
Patch courtesy of Matt Hanna.

Modified ExtractIlluminaBarcodes to handle dual barcodes.  Metrics file contains barcode read bases separate by slashes, barcode.txt files concatenate all the barcode read bases with no separator.  Removed BUSTARD_DIR option, which has been deprecated in favor of BASECALLS_DIR.

Change to MarkDuplicates to allow it to accept and merge multiple input files and write out the merged and marked results to a single file.  Merging isn't as fully-featured as MergeSamFiles.

Small change to make decoding attributes from a BAM file previously written through the SAM-JDK more efficient.

Added specialized implementation of addLocationInformation() that doesn't rely on using a regular expression so long as the default regular expression is specified.