Well, it needs to be said that a 1MB version of the Matrix will require that CPLD I talked about earlier, which would replace the functionality of the PIC, BTW, with DIPs tied to a few of the the I/Os... this is essentially what is done on the LPC 1MB mods that are out there.
A better design would probably be a generic pogopin adapter, with perhaps a simple alignment program on a PIC and a connector for hooking up a variety of... oh, never mind.... Team Xecutor plans on selling those.
As abrasive as unregistered is, he is mostly right, though. Pogopins are best left to those who can deal with volumes and design a professional PCB. If you have these kinds of resources, great, go for it, but even just a basic spin of a board can get pricey for an individual. Do you have PCAD? Can you fab your own boards? Besides the pricing issues, pogopins require precision placement (particularly the D0 pin), so slapping these things into perf board is akin to eyeballing an invitro fertilization with a magnifying glass and a knitting needle.
So you have some high hurtles, as an unexperienced indiviaul with your goal of creating a "1MB Matrix" because:
1. The simplicity of the base "Cheapmod" design relies on a specific Flash chip, which is no longer produced and is limited to 256kbytes, 768kbytes shy of your target capacity.
2. Using the larger chip will require a CPLD to properly interface to the LPC port. This will require VHDL or Verilog programming, and an in-depth understanding of microelectronics.
3. Pogopins, expensive in small quantities and requires a professionally prepared, precisele laid out PCB board to work properly.
Cheapmod works because it basically involves a cheap chip ($4), a few common components (resistors and caps), and a few wires soldered into the PC - but the work involved is probably enough for some people to just go ahead and get a Matrix and be done with it....
...and your proposal will undoubtably be much more expensive from the outset based on parts alone, and given the goodies you want to put in, it'll probably take ten times longer (or even more time than that).
I'm not trying to be a downer, but just looking at this realistically. If you have the resources to do it right, go for it.... I do know there was some guy selling nice little PCBs for homebrew Xbox modchips; he might be able to help (I don't have the URL, though). If you could get a decent supply of pogopins and a pro designer (that PCB guy, for example), then perhaps that could be joined with an inexpensive CPLD and flash.
Whatever you decide, good luck.