That sounds very odd, and I'm pretty certain that BS7671 (17th or 18th) doesn't say anything like that. I'm not even sure I've seen anything like that (as a reg) even in relation to singles, although I presume the practice then is to use conductors all of the CSA.
What BS7671 does say about CPCs (by implication, in any situation) is that although the lazy can use Table 54.7 (and therefore end up with the CPC having the same CSA as the line conductor,at least for conductors up to 16mm²), it is also permissible (better?) to utilise an adiabatic calculation, which will often end up indicating that a CPC smaller than the line conductor is acceptable - and that calculation takes no account of what CSA the line conductor may have.
At first sight, it seems to make little sense that the requirements for the required CSA of a CPC should vary according to how the cable is routed.
If there were a reg saying that it is not acceptable to put T+E into some sort of tube/trunking to protect it from mechanical damage in a 'hostile environment', then I think I would have to say that the regs were being more daft than usual! Indeed, there are regs requiring cables to be protected from mechanical and environmental damage - if that can't be done (with T+E) by putting it 'inside/behind something', how on earth can one comply with that reg?
Kind Regards, John