We are told to test for figure of 8 wiring and list it as a fault. However I have failed to find any thing in the regulations which says you can't other than "to provide reasonable sharing of the load around the ring" ref 433.1.5 which in turn directs to 433.1.1 which talks about overload.
What is described is not a true 'figure-of-8' circuit in the normal sense. If it were, then I think it's true to say that converting a ring into a (true) 'figure-of-8' will always tend to
improve 'sharing of the load around the ring', as well as potentially reducing EFLI and VD, and increasing redundancy. Hence, AFAICS, a true 'figure-of-8' is, in most senses, 'better'/'safer' than a conventional single ring, the only real 'problems' being in relation to more complex testing and fault-finding.
In the OP's case, if the two connection points between the 'primary' ring and the 'secondary' one are a reasonable distance apart (on the 'primary' ring) and/or are fairly distant from the ends of the primary ring, then the situation will be much the same as described above for a true 'figure-of-8'. As Jackrae has said, a potential issue would only really arise if the two points of connection to the primary ring were both very close to one another and also close to an end of the primary ring - but the same would be true if there were a lot of sockets installed very close to one end of the ring.
Kirchhoff's laws become complex to work out, I did in in University, but as to working out on back of fag packet forget it. I would have to get the text books out, or use electronic workbench, which is not installed on my PDA, so I would not be doing in in a private house.
As mentioned above, any such calculations would inevitably demonstrate that
any sort of 'cross-connections' in a ring will (for any given arrangment of loads)
improve current-sharing in the conductors of the circuit, as well as reducing EFLI and VD at many points in the circuit.
To actually get any absolute numbers out of such calculations would require wild guesses as to 'what loads might be plugged in where'. If, even with a simple ring, you considered the 'worst case scenario' (with a a full 32A load plugged in very close to one end of the ring), you would very often conclude that the CCC of a cable with CCC=20A could be exceeded (although much less likely if, as often, CCC=27A).
Kind Regards, John