I disagree with the situation, but it is what it is.
The reg in BS7671 is the one that says you must take into account manufacturers instructions.
The reason that they say you cannot use alternative breakers is that their DB's are |"Type Tested Assemblies" or "Partially Type Tested Assemblies".
Thus to meet the requirements of the LVD, they have to be tested as an assembly.
Thus they are only prepared to do this with their own product.
Even though there is a standard for breakers, it is sufficiently open, not only to have dimensional differences, but also, detail differences in performance & operation.
One guy I spoke to high up at Schneider, related some test results to me where by there was a significant affect in discinnection time on an RCD when they ran a simulation test. A 40A load through the adjacent MCB to the RCD varied the disconnection time of the RCD considerably. This was not the case with the load located a few ways down the board.
Remember, "they" are guaranteeing the board meets requirements, not you.
The same with pattern car parts, the manufacturer is guaranteeing compliance not the garage fitting it.
However, in this scenario, remember the manufacturers will try to lump it on you rather than take it on themselves.
So there is a slight difference.
You will find somewhere in all manufacturers data, and if not then, if they are a member of BEAMA, then BEAMA have issued on their behalf a statement saying basically you can't mix breakers & boards.
If you decide to then ensure that you have suitable indemnity, not liability insurance to defend you in the event of an issue.
BEAMA Technical bulletin here:
http://www.beama.org.uk/download.cfm/docid/F24BF81E-39AC-41C6-9557E79834520533
I don't like it either!