That's true, but it doesn't really address the question. If the cracks are a result of flight stresses, test stand burns won't reveal them.

If the tests don't simulate flight stresses, they are bad tests.

It's simply not possible to test that. This much is obvious if you've thought deeply about the nature of rocket testing.

You MUST fly to _really_ test.

Is there a good way that you're aware of to simulate the acceleration experienced by a rocket?

You're talking about a turbopump spinning at several tens of thousands of rpms, at very high temperature. Yes, a couple g's of lateral acceleration adds a bit too those stresses, but it's a fairly minor effect, I'm sure.

I think it's vastly more likely they're cracking from the rotation speed, temperature, and pressure.

I expect this is true, but I'd like to have a whole lot more information than the article contains. Specifically, under what conditions the cracks are observed.

The cracks were observed during ground testing, several years ago.