Just as an aside, I've been looking at those projects very seriously because of an idea I've had for a homebrew 6502-based computer, and unfortunately the conclusion I've come to is that while the Propeller works as a glue/peripheral chip like that it's non-trivial to achieve "cycle-accuracy" with it. (A link to add to yours would be the "Propeddle", which is essentially an enhanced version of the 6502 laptop design targeting Commodore PET emulation:
http://www.savagecircuits.com/forums/showthread.php?677-Propeddle-Software-Defined-6502-Computer
) Anyway, if you dig into the "pPropQL" design theory you'll see that, for instance, interfacing the Prop to a 68000 as "memory" requires the insertion of a substantial amount of wait states. The Macintosh OS as written is so timing dependent I don't know if you could really get away with it. (At least if you're attempting to do things like use "real" floppy drives or the serial controller. If you're emulating those devices then dealing with the "clock skew" would probably be doable.)
On the subject of emulating the IWM I'd probably suggest it's sort of pointless to include a 19 pin floppy connector, if only because the original drives are getting so rare. (And troublesome in their old age.) Is the ultimate target of the design to produce a board that's physically compatible with the original PCB, with all the same ports in the same places and allowing the use of all the original parts and peripherals? (Of course, that would either require the ports be carried on a daughterboard or that multiple versions be produced to fit either the 128k/512k and Plus cases, respectively.) I'd worry about the ultimate cost of something like that, as when I last checked the commercial version of the Atari-ST-in-an-FPGA board that includes legacy ports was nearly $600.