I'm not saying it's the case here, but there is a thing called confirmation bias where if you have an opinion or view you tend (and it's typically unconsciously) look for evidence that supports that view and disregard evidence that goes against it.
e.g. point 5, I believe every fitbit uses a 3 axis accelerometer to detect motion and it will then look for movements that are vaguely like the motion it expects from a person walking and adds a step. These are by no means perfect (the newer models are a bit better), here is an example of what one will register as steps:
Nothing is mentioned about how many etc, it's not unreasonable to expect it to measure the odd step or two as her motion from entering the water slows down, I wouldn't be surprised to be honest if the last step(s) to be measured is when she went over the weir.