Forget about rule 56 for now. Traversing would be a bad move if you know that allowance could result in an invalid patent.
1. You can't be serious. You should (virtually?) always traverse any improper rejection. If there is a proper rejection, let the examiner make it.
2. Also, you say "if you know that allowance could
result in an invalid patent." Any allowance could result in an invalid patent. About 50% of litigated patents are proven invalid. That can't be the standard for whether you accept an allowance.
The question here is whether to disclose. Although the Fed. Cir. hasn't decided Therasense en banc yet, practitioners are still separately bound by rule 56 (you can't "forget" it), and it's an interesting question whether the applicant has to submit the International publication now. For example, suppose that the examiner's rejection is wrong (even if the ref is prior art), and doesn't present a prima facie case of obviousness, then the applicant probably doesn't violate current rule 56 by failing to disclose, but probably would violate old rule 56. There's also the question of whether the Inter. pub is cumulative to the references already of record, and/or whether the Examiner can be reasonably expected to know about and review the possible 102a rejection.
Most likely, traversing would involve taking on the expense to perfect priority just to get a new office action that does not advance prosecution in the slightest even though you knew ahead of time that it was time to amend the claims. Hopefully you told the client that such was likely to happen and he doesn't think you are an idiot.
As I explained above, the applied U.S. ref does not appear to beat even the U.S. filing date of the OP's application, so I do not believe that he has to perfect priority. Also, it would advance prosecution if the examiner doesn't make the possible 102a rejection.
I've seen situations where the reference in question was the applicant's own patent or application. I believe traversing in that case and would probably be fodder for a charge of inequitable conduct. It is likely that you will not be able to convince anyone that your client didn't know his own reference was a 102(b). It's questionable enough in cases like the one john122 describes.
It's an interesting question for both post-issuance inequitable conduct and pre-issuance compliance with rule 56.