So for the desktop computer, the TE0725 isn't really an option, because there just aren't enough pins free for everything: Just the VGA output and expansion ports alone would use them all up. Then we have keyboard, the internal expansion connectors etc. Routing would also require at least 6 layers (which we know from laying out the hand-held/phone version as part of the student projects), so the board would not be any cheaper to make. It really is just easier (and cheaper in practice) to have one fully integrated board.
As for it being a "hands off sealed box", this will not be the case. There will be a number of standard PMOD connectors internally that will make expansion and tinkering quite possible. Also the cartridge port allows for much more sophisticated cartridges than on the C64.
For compatibility, we are not yet at the point of running demos to test. There are a few reasons for this. First, the C65 basically can't run any demos, because of the CPU instruction differences and signfiicant timing differences in the VIC-III. That said, we are working to improve the compatibility as much as is practicable. I'll likely post a blog post soon describing the real-time debug facilities I have added that let us monitor the instruction stream and significant VIC-IV events, such as raster line advancing and triggering of badlines. 100% C64 compatibility is not our primary goal at this point in time, only "good enough" compatability, so that we can get it out the door and into your hands. "Good enough" to me, means that it can run most games and GEOS without significant problems. But as it is FPGA based, nothing will stop us (or you the community!) working either improving the C64 compatibility, or making a dedicated C64 bitstream to deliver 100% compatibility.