Interesting topic Cammer. I've thought about this diversion a number of times in the past and although what you propose can be engineered, I think it would end up be significantly more work and $$ than one might imagine. For a long-term solution, the riprap used for the diversion would likely be much larger than what is currently placed along the left bank by the FN facility; this is because of the curvature of the diversion berm and that river flows would almost be directly impinging on it. Some sort of impermeable feature may also have to be constructed along the downstream side of the berm to prevent river flows from trickling through the riprap. Depending on how robust the structure needs to be, the riprap may also need to be keyed into the river bed to counteract scour and undermining of the berm. All constructable...
What I think would be the biggest hurdle would be the regulatory requirements for a diversion of this magnitude. Even small diversions typically require salvage of any fish in the portion of the river that will be isolated. Imagine salvaging any resident and migratory fishing along a reach of river that's over 2 km. You mentioned that the river was diverted once in the 80s. I'm not familiar with what was done at that time with regards to the size of diversion and environmental requirements, but suspect that it wouldn't fly now a days. Though I may be missing something and would be interested to hear what others think on the subject.
One other thing is that the avulsion point of the river is on the fan of the river. From a geomorphological perspective, this is the reach of the river that would tend to experience net aggradation of sediment over time, which would be the main driver for avulsions here in the past. By constructing a berm at that location, a large portion of the fan would be cut off leading to increase aggradation elsewhere and potentially unpredictable behaviour of the river in the future. I admit that I'm not aware of all of the issues facing the hatchery at this time, though I'm generally in favor of letting the river migrate naturally and modifying the hatchery channel to suit. If flows along the hatchery channel are an issue, perhaps designing an intake system to increase flows into the channel a feasible solution? not sure what's in place now, and that would be costly as well but more predictable. Thanks again for starting this topic.