The inputs and outputs for components are actually implemented quite nicely in DLS. I haven’t tried modifying them myself, but from a static code analysis, it looks like just adding the appropriate new ChipType to the ChipDescription method should do the trick. Most of the magic happens right at the beginning of the SimChip method in Assets/Scripts/Simulation/SimChip.cs.
Adding a larger ROM size shouldn’t be too painful either. The UI editor for it is written in a very elegant, clean way, so it should theoretically accept data arrays of any arbitrary size without breaking the layout.
The real question is how the simulation engine itself will handle the performance overhead once you start cycling through that much data in real time. ;)