I wouldn't assume, just because the switch has the same part number, that the evolutionary (internal to the switch) changes aren't traceable.
It;s a wonder nobody has sued for a car that can;t run out of gasoline. Same ultimate failure mode as far as vehicle control goes. Engine quits, Armstrong Steering becomes effective, etc. I can picture a designer's judgment that the unintended shutdown is a manageable event.
... it is possible that he did not examine any GM documents related to the part. I do not think such docs would be proprietary. If the engineer were part of an ongoing lawsuit, then I think the discovery process might have made these documents readily accessible to him. Maybe (for one reason or another) he just didn't view the engineering drawings. But it sounds like he sure was scratching his head wondering, "What's going on??"
But if the engineering documents were properly maintained, they should have easily shown that revisions to low-level pieces of certain configuration items were made. It shouldn't have been hard to ask, "Why'd you change this part right here?"
It's an assumption on my part, but it seems reasonable that the change was made by GM and that the documents did not adequately reflect the change in form, fit or function. Either from sloppiness or malfeasance.
High-level identification (serial number, part number, what-have-you) may not tell the whole story, but the revision history should never be a secret or a mystery.