What we mean here is neither "an object" (something that exists at runtime) nor "a class" (a type), but rather "an inheritance":

struct A {};
struct B : A {};
struct C : A {};
struct D : B,C {};
int main() {return 0;}

In this program there are 0 A objects, 1 A class, and 4 A inheritances (2 of them direct).

So the proper fix here would be to introduce the whole new term/concept into the standard?

That, or else explicitly state that "base class subobject" has a meaning as a component of a class as well as as a component of an object. This isn't entirely unreasonable: after all, a complete object of type X has exactly one of the object-kind for each of the class-kind that X itself has, and it parallels "class member"/"member subobject".

This got reworded by P1787R6, applied by commit c7e8927
This issue seems to be still open, though.