Everyone is being fooled by the appearance of the plug and socket, due to familiarity with the socket outlet as a means of connection for portable appliances. The plug here is not part of a portable appliance, so the dedicated plug-and-socket arrangement becomes a piece of switchear along the run of submain from main DB to conservatory DB. Consider it to be equivalent to an FCU, but using withdrawal of the plug to achieve double-pole isolation. No-one would argue that a permanent circuit fed from an FCU is not part of the electrical installation, surely?
Then, the question is whether as a piece of switchgear, the plug-and-socket is adequate for the purpose. Is its current rating high-enough for the design load, for example,? What about the COP guidance that limits the BS1363 plug earth pin to a maximum design leakage of 3.5mA? You would not really want that limitation on a submain. Used as an isolator, it also switches the CPC, which it is not supposed to do. Overall, I see this as an inappropriate piece of switchgear on the submain and would code it according to its shortcomings in that role.
For anyone still convinced that connection via a plug excludes the conservatory from the installation, consider a theatre-lighting system where all the circuits are patchable. Every dimmer channel (=DB way, which might number dozens or hundreds) terminates in one or two socket-outlets, often directly adjacent to the DB. Every installed circuit cable leading off around the building begins with a plug that can be placed in any of these outlets, like a telephone exchange. Would you then consider that none of the installed wiring is part of the installation?
Note that in the theatre patch panel, the CPC of all the circuits connects to the MET so that it does not rely on the contact of the plug and socket, and remains connected with the plug out. The plug and socket CPCs are also connected for completeness.