v1.0, 2014-01-04 -- release
v1.0, 2014-01-15 -- fixed a minor bug 
v1.0, 2014-04-09 -- added to github
v1.0, 2014-04-10 -- fixed minor bug on analysis.py regarding output file names
----------------
v1.1, 2014-05-01 -- Now can compute N(z) and True plots using output PDFs
                    Check use_pdfs in utils and plot_pdf_use in plot
                 -- Add multiple file options to write PDFs files per processor
                    which is faster for BIG files
                 -- fixed bug on multiple output file names
v1.1, 2014-05-02 -- Added PDF storage module
                 -- Add plotting routines to plot sparse representation
                 -- Add keys to input file
                 -- Optional fits output (original PDFs)
                 -- Updated documentation
                 -- v1.1 Release
v1.1, 2014-09-25 -- Output inputs file fixed
                 -- Update usage prompt
v1.1, 2014-10-16 -- Read fits files as input (pyfits 3.3)
                 -- Added argument options (--no_train, --no_test, --no_pdfs)
                 -- do: ./runMLZ --help for more info
                 -- Added colored outputs
v1.1, 2014-10-23 -- Added --check_only option to check the code before a big run
                    It only train and test on a subsample of the data
v1.1, 2014-10-25 -- Added --print_keys (-pk) option to print current values of keys
                 -- Added --modify (-M) to modify on the line command values of keys
                    e.g., name of the file and range of z, --modify minz=0.1 maxz=1.2
                    testfile=data.fits
                 -- csv catalogs supported
v1.1, 2014-10-27 -- Added --replace (-r) option to replace additional output files
v1.1, 2014-11-07 -- Change sparse representation storage, now first element stores the original amplitud
                    instead of 1. (for block summations of PDFs, 2 steps are needed)
----------------
v1.2, 2015-04-08 -- v1.2 release, faster than previous versions, fixed several issues
                    See previous changes
