I've been trying hard to think how they could have done this better, in a matter which was rational, logical, consistent and 'sensible'. It's not easy. It might even be essentially ';impossible'!
I think I have now convinced myself that there is no rational or logical way to attempt to make a distinction between OPDs on the basis of what 'enclosure' they exist in - so we can't really have a (notifiability) rule based on a BS7671-like definition of a 'new circuit' which does not 'catch' FCUs.
We could therefore start by defining a (notifiable) 'new circuit' as one originating from any OPD which was either new or previously unused. That would obviously 'catch' all FCUs (as well as circuits originating from, say, a 'secondary' CU in an outhouse) and if we wanted at least some uses of FCUs (or, to be logical, any equivalent 'OPD with In≤13A in an enclosure') to be non-notifiable, then that would have to be dealt with by specific exclusions. Exclusions one might want to contemplate would probably include fused spurs/branches from a ring/radial final and FCUs (or equivalent) supplying a single fixed item of equipment (perhaps including a single light fitting plus switch).
However, 'sensible' though that might sound, I think I would seriously struggle to justify those 'exclusions' rationally. If we felt (as we do) that a light+switch fed from a (new or previously unused) B6 should be notifiable, how could we rationally justify saying that feeding it from a (new or previously unused) 3A or 5A FCU should be non-notifiable? Similarly, if we feel (as we do) that installing a number of new sockets on a radial originating from a B16 is notifiable, how could we rationally justify saying that adding sockets to an existing ring (or radial) final via a 13A FCU (i.e. 'a fused spur') was not notifiable?
I fear the fact is that none of us (not just the authors of the Building Regs) has a very clear, rational and consistent idea of what should, and should not, be notifiable!
Kind Regards, John