Consider the setting \(E(Y)=Xb\). A linear function of \(b\),
say \(l'b\) is estimable if and only if there exists an \(r\) such
that \(r'X=l'\) or equivalently \(l=X'r\). Hence \(l\) must be in
the column space of \(X'\), i.e. in the orthogonal complement of the
null space of \(X\). Hence, with a basis \(B\) for the null space,
is_estimable() checks if each row \(l\) of the matrix \(K\) is
perpendicular to each column basis vector in \(B\).