Better observability on parse rejections

[rcgoodfellow]

When a parser stage rejection happens, we need better observability. Ideally, we would have a USDT probe that shows the complete set of parser state transitions that led to the rejection, including the specific rejection point. Right now, one can only look at the packet that got rejected and try to reason through the state machine code manually.