I think their approach of understanding how the protocol actually works and making a latency tolerant networking layer on top of it might work for pokemon (assuming the pokemon game code allows this)
Imagine the game sends a message over the wire and presumes it can get an stateful ack calculated by code running on the other side before it takes any other action, and then does that multiple times per frame. Also, the code might not even block while waiting arbitrarily long, it might just run ahead and desync/bug out if the reply doesn't come back in a couple ms.
What do you expect the emulator to do when the network latency is more than 25ms long? The emulator can't trivially simulate 0.1ms latency instead. It can maybe try to hook in and pause the emulation while it's waiting for the reply, but it would grind down the game to a crawl.
This is a technique used by some fighting games that have internet play, and is also used in emulators to deal with input latency on modern TVs.
Coming from the ROM hacker scene I initially was confused by this; too - but absolutely understand it’s impracticality difficult from a technical end.
edit: okay looks like their scheme doesn't work for arbitrary games and arbitrary protocols.