Again, the Scalix Gurus(tm) here can speak better to this, but in general:
1. The Scalix mailstore is located in /var/opt/scalix/??, where "??" is usually the first and last letters of the hostname. Ex., mine is "pop," so my directory is "/var/opt/scalix/pp." The posts here usually refer to it as "~/" as a shorthand.
2. The "~/s" subdirectory is where the maildata resides.
3. The postgresql stuff (in the directory "~/postgres") is simply an *index* of the mail data. That's why you can delete it and rebuild it. (But that's also why the Container server has to be running when you do.)
4. All of the mail messages are actually stored in ~/s/data. You can go into those folders and see the actual messages, complete with headers and "to:/from:" stuff.
5. [edited -- added] The ~/s/user directory contains per-user data. You'll see posts here about people having imap and login problems; you can delete the per-user imap-cache in this directory and let it rebuild to solve those issues.
(BTW: on my system, I actually have "user," "user01" and "user02" directories. This is an older mailstore, and with fires and hires over the years, I guess we've grown. I haven't seen anyone else here mention that, but it intrigued me the first time I saw it.)
The directions in the link that I gave you about rebuilding contains everything you need to backup and restore. If your installation is as blown as you say, and you can reinstall Scalix, that's a good idea. But you might want to check your system (including the hardware) to find out why it should blow up like that.
If you're rebuilding from scratch, this would also be an ideal time to change OS, too. You didn't say what you're using, but I've had good results with Opensuse, and *great* results with CentOS. The post that I linked to? We had finished our CentOS rebuild about a week before that date, and it has been running without a glitch since then. Not even a burp.
www.centos.org -- we're using CentOS 5.2.