Version 2.1 (beta).
Created on 11/14/03.

Changes from 2.0:
1. Default parameter specification removed from parameters_h0.cpp and 
   parameters_h01.cpp, since such specification is already present in the .h
   files. 
2. lib/newmat recompiled.
Both changes above were required for compatibility with RedHat9.0+
3. directory mlagan/ added in lib/. This is lagan 1.21. README changed to 
   reflect addition of lagan.
4. A bug-fix in helpers/extract_lagan_blks.py

Changes from 1.1: 
1. FringeCorrectionFactor bug corrected in parameters_h0.cpp, parameters_h1.cpp 
2. README changed to explain stubbh01 
3. Makefile changed to allow _WTMX_BIAS

Changes from 1.2:
1. stubbms and stubbms_modifyalignments changed to:
   in case where a homologous window has some pseudo inter-block fragment
   larger than 2*windowsize, only that fragment is dropped, rather than 
   dropping the entire homologous window.

Changes from 1.3:
1. Provision for two new programs: stubb_fitprobs and stubb_fixedprobs
to allow for stubb without parameter training.

Changes from 1.4:
1. README changed to explain stubb_fitprobs and stubb_fixedprobs
2. gsl dependence removed. all matrix functionality comes from newmat
3. mlagan removed from lib/  -- it has to be obtained separately. README 
   changed to explain its installation and integration with Stubb.
4. dialign support removed completely.
5. helpers/prepare_for_stubbms.pl translated to python, so that all scripts
   in helpers/ are python scripts, for uniformity. Of course, lagan still 
   needs perl.
