If they did actually fit correctly, and were the correct specification, then in spite of not being type-tested there would not be a safety issue
That much is apparent. However, are you saying that they would still have to be regarded as not being 'type tested', hence not a "CU" per BS7671 definition?
If the published relevant dimensions were identical for both switches, would you feel that whether or not it 'fitted satisfactorily' would still be "a matter of opinion"?
If the tolerances were included, and the materials were the same. The reason we type-test is to verify things, so that opinions don't matter.
Without a ludicrous amount of testing (involving a wide range of {'within tolerance'} actual sizes of relevant components), the issues of tolerances cannot really be 'verified' experimentally - that really has to be done on a theoretical basis, doesn't it?
I do wonder if if we're not at risk of losing sight of common sense. If, in 'everyday' terms, the component 'fits', then I would suggest that the way in which the component is installed (terminal tightness etc.) is a
very much more important factor (in terms of safety) than are such issues as dimensional tolerances and materials.
We've been over all the theoretical possibilities in the past, but I wonder whether there has ever been a real-world case in which it has been proven that some sort of 'failure' of a CU was the consequence of it having been populated with one or more 'wrong make' (but 'dimensionally similar') components?
I wonder what is your view about chopping bus bars? Who is to be the arbiter of whether or not it has been done 'satisfactorily enough' not to invalidate the type testing?
Depends on the MIs. If done in accordance with them then the person doing that has to make a judgement. If not, then no, the type-testing is invalid. Decent manufacturers will have tested various configurations of their products.
If we are being 'pedantic', then the most relevant factor is surely the way in which the cutting is done (e.g. whether it results in any twisting, or other distortion, of the bar) and there's no way that a manufacturer is going to know, and test, every possible variant of that.
Kind Regards, John