Yes, I am saying the same as you. Why, then, when a 30A BS3036 (20A/0.725 = 27.6A) fuse is replaced by an MCB is it not allowed to use a 40A MCB or - ...
I assume you are talking about ring finals. For the reason you imply, it has never made any sense at all to me that 433.1.103 allows a ring final protected by a 30A BS3036, when it restricts one to 32A with an MCB - as you say, these are not consistent. ... you therefore would have to ask the authors of BS7671, not me!
Why, for new ring finals, do we not use 1.5mm² conductors?
The short answer is, of course, that the regs say that you mustn't, unless it's MICC. As you imply, provided one uses Table 4D5 (and not 4D2A!), or singles (Table 4D1A), the CCC of Method C 1.5mm² is 20A - so why 2.5mm² cable with a CCC of 20A is allowed, but 1.5mm² cable with a CCC of 20A is not allowed is, again, something you would have to ask 'them'
I suppose they might be sceptical about whether people always pay proper attention to installation method and de-rating factors, and therefore are being a little cautious/'pragmatic' - with 1.5mm², anything short of pristine Method C throughout the cable run would bring the CCC below 20A, whereas with 2.5mm² it's far more likely that the CCC really will be at least 20A
As the 0.625 ( 5/8 ) is only just over half what advantage (today) is there in installing ring finals at all?
Well, that's obviously a whole separate debate, and I need not tell you that some people see no point in, and/or don't like the idea of, ring finals. I'm personally pretty neutral, although I am not all that keen on the idea of 20A multi-socket circuits, since that theoretically can't even support two 13A loads. However, I'm perfectly happy with 32A radials as an alternative to rings.
Kind Regards, John