IMO the answer will eventually come back to the overall risk assessment as used for the hazard analysis.
IMO yr first MD could be considered as currently screening the nuts (at a certain MD sensitivity level).
You then have a further, final stage where you have to assess the total risk of metallic contamination.
If screening is considered Ok as per yr target requirements the net risk will presumably depend on final stage only.
If the risk at the final stage is (a) considered significant, this stage will IMO then be the CCP. If not, then (b) the screening stage MD will IMO be the CCP.
If (a) the critical limits may then be considered the detection sensitivity of the 2nd MD, similarly if (b) >>> the det.sens. of the 1st MD.
To be satisfactory, the respective sensitivities will both need to meet any regulatory/customer requirements.
(Of course, if you go for (b) an auditor may ask you why you insert the 2nd MD. = Just for QA purposes perhaps ? )
This is a rather non-simple hazard analysis IMO, other people may have different opinions. For example i have seen textbook cases where a preliminary magnetic screen was also selected as a CCP in addition to the final MD/CCP with the justification that the 1st screen acts as an important ("critical") assist to the final MD. So you then get either 2 CCPs or a "combined" one. It's justifiable within haccp.
PS - with respect to oprp's the decision should logically depend on yr risk assessment/hazard analysis/interpretation of iso22000 clause 7.4.4 (a-g)/iso22004. There are many, many, equally auditorially acceptable, viewpoints over selecting oprps. Based on threads on this forum, the auditor typically wants to see a logical selection procedure for ccp/oprp, not so much details about the choice of the method. And how you validate/implement it.
Hope the above is intelligible.