Also, as you say they've built a prototype to test the result. To be taken seriously, they should have also put a few students on the task of building a similarly sized Kaplan turbine and then compared the efficiency. It would have been very simple; there are probably even scale model Kaplan turbines that you can buy for this purpose. (It's basically just a propeller in a tube.) TFA is a bit like an SW dev coding up a new webserver, claiming it solves a case where existing ones do poorly, but then not bothering to benchmark it on that case against, say, Apache or Nginx.
Finally, the big challenges in hydropower are not in increasing efficiency or lowering cost. The challenges are that we've interrupted/dammed most large rivers already, and any new plans to build hydropower face great resistance both from NIMBYs and environmental protection groups.