Come on, what wrong with the current fdisk/bsdlabel? If you need support for a new FS, just add it to the currently used utilities.
What's wrong with the current fdisk/bsdlabel is that it doesn't support any other disk layouts than the basic "BSD partitions in a slice on an MBR disk". Given the number of things FreeBSD actually supports (gjournal, gmirror, graid3, GPT partitioning, geli encryption, ZFS raids, atacontrol raids, vinum raids, booting from several of the previous), that's rather limiting.
As for extending sysinstall to support these things, that's definitely possible - but not by extending the current partitioning interface. (You can stack GEOM modules and RAID disks in ways that don't really fit into the "split free space into chunks"-spirit of it.)
And for those two things (so far...) running in "graphic modes", what do you propose? That FreeBSD's base installation include and require Xorg? What's next? Ding/Ploc/Vlam-type sound schemes, transition effects, etc. Ã la Windows?
There's absolutely no correlation between what the installer uses and what has to be in the base or installed system. You could build a live CD with an installer, KDE4 with 3D acceleration, a large sound theme, and an automated pet walker - and then use that to install the same minimal installation you can get with sysinstall today.
Heh, you could in theory install FreeBSD from a Linux environment, if gpart and their UFS2 tools are up to it. I'll have to try that some day.
Having an optional X interface to the installer gives some benefits, like making it easier to present information (higher resolution, graphics), the convenience of mouse support, and better unicode handling. Not must-haves, but not useless, either. Running X from a live CD is a solved problem, so if FreeBSD creates an installer based on a live CD - image, using X to make life a touch easier when possible seems sensible.
Oh, and there is a related
project going on already. It was a bit lacking last time I tried it, but maybe it'll grow up into something nice.