NDPhA, while an important argument for the potential non-CoC positive results of the Ar-NN=O substructure due to transnitrosation (or indeed denitrosative formation of a hydroxylamine, following the aromatic amine pathway subsequently, as we describe here Tennant et al, Ponting and Foster - we only found this after writing the referenced Ponting et al), isn’t quite the right argument for the nitrosoindoles.
Should note that it is for the nitrosoindolines, where the N-containing ring is not aromatic, and these can be classified as CPCA and will be class 4 or 5 depending on whether the 3-position is substituted (class 4 because (0,2) and in a 5-ring is 5 points, you’d need to add both activating features to get to less than 4)
Returning to nitrosoindoles: As you note, they are positive without S9; this is because the pyrrole-type nitrogen behaves a lot like an aromatic carbon (with a bit more electronegative character). The relevant analogues for nitrosated pyrroles and indoles are therefore C-nitroso aromatic rings (such, ironically, as that formed by the transnitrosation of NDPhA, though not of course NDPhA itself where the amine is not in an aromatic ring), which reduce to the hydroxylamine via innate bacterial metabolism and follow the aromatic amine pathway from there. They should as far as I am concerned be treated as normal, non-CoC, potentially mutagenic impurities, though do flag them as an N-nitroso compound you have thought about rather than hiding anything…
Ultimately I think the HAs haven’t been particularly clear with their nomenclature here; I would say pyrroles and indoles are not amines and thus don’t form nitrosamines per se, rather ‘N-nitroso aromatic systems’