Steel fails from ductile fatigue cycling, so the cracks only grow when they use the gates. As long as they limit the number of open-close cycles they can limit the crack growth. Cracking is expected and starts long before failure happens, indeed "failure" typically happens because a part grows too large to function (because of stretching and/or cracking), not because it actually breaks. The behavior of watching cracks grow is standard procedure and works within a fairly large margin of certainty

HOWEVER

Something like this dam should be held to a much higher standard. Even a 1 % chance of sudden failure is totally unacceptable. A pipeline? No worries. Bridge? Okay, fine. Horrible if it fails, yes. Lives lost, most likely. But just the top third of this dam contains approximately 5 PETAJOULES of potential energy. Suddenly draining 2 million acrefeet from that dam would release as much energy as a 1 megaton nuke.

Extreme scenario? Yes. Unlikely even now? Probably, I don't have access to reliable information. Worth protecting against even the tiniest chance, at any cost? Absolutely.