Regarding the first disclosure, I remember reviewing the rate limiting pull request, but didn't know about the 32-bit nIdCount field.

The PR didn't touch that variable, but just indirectly caused it to increase less. An alternative fix would have been much easier: just make it a 64-bit number. But that would have been obvious to anyone looking. Since the rate limiting was useful on its own, it makes sense to me that this indirect fix was used.