Full nodes undertake a "UASF" and only accept blocks from the miners that are signaling for segwit and thus bring about a fork.
Well, then these nodes will stop if the block chain has a single block on it not signalling segwit.
That's based on the assumption that at least some of the miners that are currently signaling for segwit would not "switch" (at some point at least) to build the "segwit" blockchain.
That block will never come.
Again, you're basing that on an assumption. It will never come if all miners stop mining their current block when they receive the next one, i.e. they accept the next one no matter what. A miner that joins in the UASF will reject the next one if it's not signaling for segwit and continue mining his block.
Ah, you mean, when
the miner pools fork ? So again, this has nothing to do with Joe running his UASF in his basement, but that
miner pool deciding to fork.
We're back in the scenario where the miners fork: some build the classical chain, others build the Segwit chain. If miners don't decide to fork, like right now, whether you run your blocked UASF node in your basement or not, will not change anything. If on the other hand, miners do fork, whether you were running your node or not, didn't matter either. And you don't even need to have such a node around: you can connect your wallet to one of the pure-segwit miners, and you have your coins on the segwit chain, or you can connect your wallet to one of the non-segwit miners, and you have your non-segwit coins.
So we have now two coins: the BTC-segwit coin, and the BTC-non-segwit coin. Maybe exchanges will list the two coins. If you want to transact, you will need the two kinds of chains. But the two kinds of pools have them. No worry.
And, AGAIN, the non-mining node wasn't involved in this split: it were the segwit miners that decided to fork, and gave rise to two different block chains.
And, again, this will be determined in the market.