The files are optimised using an equivalent of cpp but for shell files. No more need to 'case $OS' into a file
Any occurence of a UNIX command is replaced by its absolute path for the specified system: allows robustness and security
Programs which can configure the scripts to fit to a file layout
Programs which can do a minimal translation from csh2sh for example
With the new versions (above 2.4.1) the scripts are compiled which means that each occurence of any UNIX command is replaced by the command name plus the full absolute path.
This is needed to prevent the following bug. Assume a user redefined the 'whoami' command and put it under $HOME/bin. If in the file sys.conf.* you set USERPATH to be '$HOME/bin /usr/bin ...' . As $HOME/bin is before any other system directory in the USERPATH variable then in the PATH variable, the scripts will call the user 'whoami' command instead of the system one and they will probably break.
Another feature of the compilation step is that the parsing algorithm which was evaluating the environment variables you provided in the sys.conf.$OS file is replaced by a simple source of a file sys.conf.csh or sys.conf.sh depending on the shell flavour of the shell (csh = C-shell; sh = Bourne-shell).
The latter files are the output of the compilation procedure and are the source of these two files is the previous sys.conf.$OS. It increases the performances and the flexibility.