It's case # B226061, Court of Appeals of California, Second District, Division Three. The decision can be read here.

The court was obviously cognizant of your reasoning why indexing could not possibly be a determinant of priority, saying, "it would disrupt the statutory scheme to make priority turn on the random act of indexing, as defendant advocates, especially where banks and title insurers have no influence over when the recorder indexes trust deeds."

There's nothing in the decision regarding L.A. county's system for detecting conflicts or producing a fallout report like you described, and so there's nothing in the decision regarding how all these mortgages went undetected.  Perhaps the pleadings in the case shed some light on that, and if you should happen to get your hands on them, I'd be interested to learn what you find out.

One thing that I suspect is that the plaintiff argued that the index timestamp mattered because that was the only criteria that was in their favor.  The decision doesn't say anything about document numbering being a factor in determining priority, but I suspect that's simply because that wasn't the plaintiffs argument.  The decision implies that there can be other factors beyond the recording timestamp which determine priority. (The ruling says that "Other things being equal, different liens upon the same property have priority according to the time of their creation".  I take that to mean, then, that if relevant other things are not equal in a case where the time of creation is equal, then the liens would not have the same priority. )