10/22/2007 - s2hat v2.0(beta) released;
10/25/2007 - mask2scan - bug fixed;
10/25/2007 - scanBcast - syntax change: on the input:
             variable of a pixelization type not a pointer to it.
01/30/2008 - s2hat v2.01(beta) released;
02/10/2010 - s2hat v.2.5(beta) released:
           - bug in map2alm fixed 
                 -- it was showing up only in the multiply-connected
                    observed sky patch cases;
           - a total number of pixels (often used only implicitly) in the full sky maps
             accepted by the library has been extended to be any number, representable
             by an 8 byte int (instead of 4 byte one);
           - disconnected from HEALPIX library 
                -- variable types now defined internally rather than via healpix_types;
                -- does not have to use any of the HEALPIX routines (but can use HEALPIX
                   FFTs if desired);
           - s2hat specific types: 
             -- e.g. s2hat_int4 (C), s2hat_i4b (F90) - all defined in s2hat_defs.*
             -- also for consistency the structures renamed accordingly:
                    pixeltype -> s2hat_pixeltype;
                    scandef -> s2hat_scandef, etc ...
           - permits a choice of an external FFT package;
           - Monte Carlo optimization of thr FFTs;
           - platform-specific optimization for map2alm routines;
           - SMAXCHNK can be changed on compiling;
           - partial map and partial alm distribution/collection routines.
02/15/2010 - s2hat v2.51(beta) released:
           - comment line problem in the -DSP3 case fixed;
           - example codes and scripts added to the tarball.
07/15/2010 - s2hat v.2.52(beta) released:
           - preprocessors comments in F90 files moved to the first columns as required by
             some compilers;
           - minor bug fixed in plm_mvals_gen;
           - bug/typo corrected in s2hat_f2c.h (thanx to Sam Leach).
09/08/2010 - s2hat v.2.54(beta) released:
           - changes in a way F90 pointers are used to make the routines work for some 
             versions of gfortran compilers (e.g., as distributed with Ubuntu, etc - and 
             likely also with other Debian-based linux distributions), which it used to
             fail before. The problems was with an assignement instruction pa => a,
             where a is an allocated array and pa a corresponding pointer. It turned out
             that if an index of "a" goes from 0 to n then a corresponding index of "pa"
             goes also from 0 to n for most of the compilers (including some gfortran ones).
             However, for gfortran distributed in ubunbtu the index of "pa" runs from 1 to n+1.
             Note that if the assignment explicitly states the index ranges than the pointer
             indices are always shifted by 1. I.e., pa => a(0:n) then the values of the index
             of "pa" are from 1 to n+1;
           - minor fixes in the way the plans are used in the case of HEALPix FFTs.
04/30/2012 - version 2.55 beta released:
           - minor bugs fixes: 
             -- in collect_partialalms (module: s2hat_toolbox.F90);
             -- in s2hat_ring_synthesis_fftw_c2r and (related) in fft_mc_setup (module: s2hat_toolbox.F90).