Starseqr

* Big release!
* Better modularity in code
* Speed improvements for all annotation functions
* Improved logging
* Fixed a bug in bed subsetting when using the "both" style
* No longer produces the same style of bedpe/VCF for Fusions.
* New Fusion bedpe using proper coordinates
* Support fastqs are broken into more coherent groupings: span, split, overhang
* Cross homology checks now use span and overhang reads separately
* Fixed a primer bug where breakpoints were not being used correctly in some cases
* Fixed several bugs that were introduced that prevented DNA mode from working.

* Updated to work with pandas 19.2 and numpy 11.3
* Fasta index is created if non-existent
* Candidates.txt shows PASS or fail reasons

* Changed all realpath statements to abspath.

* Change absolute path requirements for gtf and fasta to accomodate docker

* Added a script for making a DREAM challenge compliant bedpe

* The smallest portion of a split read is now used to calculate number of frags >20 as MINFRAG20 or MINFRAG35
* More filters are applied to remove FPs with certain read characteristics such as diversity, MINFRAG20, and Cross-Homology
* Candidate info has been converted to not contain nested lists
* New output format with more info