I haven't had time to look at those saves to see if they match what I was given. Ajxpk is right about the trash bytes. The original trash bytes are generated from the actual event process. I wonder if that process was modified (Like a program that has a generate 1, 5, 10, 100 option) to account for the other 4 mews in the 5-mew block. As for the trash bytes, I haven't had time to put together a pattern search algorithm. Although since it is just one byte, and there are several variables that could account for trash bytes, it is hard to start.
if I had more information about how trash bytes are normally generated in 3rd gen, including (or not) event process, then that would help tremendously. Unfortunately it could be something as simple as, what was in memory at that time, before the generation. Also if 3rd gen has dynamic memory offsets (Emerald had that too, did FR?), that could account for the trash bytes being different too.