The technology exists to find internal flaws, as has been mentioned above. Ultrasound would be one good route to go. The trouble with UT in this case is the irregular shape of the part. Round shafts are duck soup, but we already know the failure is happening at the very edge of the journal and into the throw.....that is where the hard part is. Yes there are handheld UT units that you can use on shafts and simple parts. We used one in our plant for years until we upgraded to an immersion unit.
I am sure there are people set up to UT cranks. It is FAR from simple unless you have very specific setups where you can shoot the areas of interest and not get a complete bullshyt signal back. Then you have to have a standard to estimate a flaw size based on the amplitude and timing of the reflection. The technology is easy to understand but the application can be mind numbing.
But here's the bottom line. Unless you have a crack, UT doesn't show you anything. Nada, zero, nothing...until it starts. So the practicality of finding a crank that is just starting to fail and getting it UT'd is very low. Especially when the cracks appear to be originating at the surface (which makes some sense to me). It's entirely possible that the core of the crank is at a lower stress state and finding voids or cracks there could be almost meaningless.
One of the effects that Fingers postulated that still rattles around in my brain is the unsupported load at the end of the crank. If that is one of the main root causes, alternate fire may help but not completely solve the issue.
And yes the life of the crank will definitely contribute. A brand new crank in an alternate fire setup could last longer, but you'd have to run some extensive experimentation to prove that beyond a shadow of a doubt. A sample of 1 or 5 or 10 ain't gonna be enough to get statistical proof, especially when a lot of those setups are in pullers that have maybe 30 minutes of high power run time per season....
Just throwing out ideas.