Agreed, there's lots of margin to increase preload above what Toyota recommends. Not much of a downside to it, either. And the bolt will fail in tension way before the knuckle threads will yield, provided you have 4 or more threads engaged.
Regarding bolt failures in the field, I've seen a FB post of someone having that happen, but it's not clear how they know it was a shear, rather than tensile, failure. Any chance you have pictures or remnants of failed LBJ bolt joints?
I'm having a bit of a hard time seeing how the joint can fail in shear as long as the bolts are tight. In addition to the friction force, there are two shear cones on the LBJ that should take the shear loads. What happened to these cones to let the bolts fail in shear?
My suspicion is that the bolts start to back out due to the heavy loads you describe, and allow gapping to take place. Once the joint gaps, the impact loads lead to oscillating loads in the bolts, the shear cones are no longer effective, and the bolts do in fact fail either from shear or tension.
That's why I personally think that the most critical thing in the LBJ bolts installation (aside from preloading to the right load, around 10,000 lbs or more) is to ensure the bolts don't back out. And after way too many years in the aerospace industry, I can tell you that loctite is not the best way to ensure that.