Lots of cells and tissues remain alive for months after death. It's easily possible that the nerves and muscles of this fish are alive enough to trigger basic autonomic swimming responses that are powering it upstream.
I want to see an artificial fish model that shows this behaviour.
from the paper, which was published in 2006. and re a downstream comment, they euthanized the fish with a bath of anaesthetic, also in the paper.
If you do that test where you apply electricity to a dead frog's leg to get it to kick, it will only work a few times before it's out of juice.
The energy comes from stores in the cells and when those stores are depeleated the cell will replenish the stores by digesting itself.
A fresh dead and pithed fish might flop, and might even flop several times in a row but that's about as far as it goes.
E.g. Just a change in temperature can make an efficiently swimming fish's red muscles’ duty cycles so maladaptive they are doing negative work, swimming requires really specific firing patterns.
Also consider — chances are they were moving an anesthetised fish into the experimental apparatus without realising it overdosed, in which case innervation would have been negligible.