A pretty straight translation from the ruby code in the jcoglan book.

This code allows you to read and write a single gitconfig file.  

Its a bit nasty as theres lots of string manipulation and the whole 
idea of a key (foo.bar.baz) being split into [foo "bar"] with a baz
variable and back again is all string heavy.  Also I had to create 
the various containers to allow for string arrays.  Sections are
stored in an array and a map because a map wont free the sections
but a section will etc etc.

ANyhoo, the smakefile creates a configreader binary that you can ignore
its just a fake testharness.  The real outputs are configfile.lib and
configfile.h which I copied into AmigaGit.  

Over time we can improve or replace this code without cluttering up the 
AmigaGit codebase too much.  Maybe building various parts of git in libs
might make some encapsulation sense :-)

I still have to make a makefile so this lib can build for vbcc.  For the moment the
VBCC build of AmigaGit is broken as VBCC doesnt like the SAS-built lib.