Sounds like that NEO-G0 is getting enabled when it shouldn't be.
https://wiki.neogeodev.org/index.php?title=NEO-G0
If you look at the internal logic section it shows that pins 39 and 51 are what could enable parts of the program rom to pass to the CPU data bus. I would check to see what those pins are doing.
Also, In the lower left of the page it has the pinout for J4 on MV2B which should be the same as MV2F. In the comments for that it gives the impression those enable pins come from the NEO-I0 chip.
Your MAIN RAM TEST likely just means you have all your hard dips in the ON position. You also seem to have a 32K/slow vram related issue given the dupe/corrupt text.