Shorten the tag used for haplotype jobs from 'haplotype' to 'hap'.  Because it screwed up formatting in the resource table, grrr!

Add merylOperation::finalize() to clean up what we can - like the input buffers - when we're done with the operation.

Fix error in computing thresholds at small frequencies; remove dead code from unused high-end threshold.

Accept a haplotype noise threshold directly on the command line, or compute it from the reads histogram.

Make smaller overlapper jobs to prevent more than one hash table build per job: increase overlapper hash load limit from 75% to 80%; decrease base partitioning limit scale from 1.333 to 1.25.

Save the splitHaplotype log and use it to decide if the unassigned reads should assembled, or even if the assemblies should be started at all.

Update overlapInCore threshold parameters to adjust for meryl's automagic filtering of unique kmers.

Fix stopAfter for meryl; add meryl-configure, meryl-count, and meryl-process to the existing meryl stopping point.

Do not return from nextMer() if the kmer has a zero count; move on to the next mer (unless it too has a zero count, in which case we'll move on to the next mer (unless ....)).

if not present, report version as snapshot (issue #1111 ).

Make haplotyping (mostly) work in cloud mode.  It still requires input files to be on a network disk.