No, $\mathbf r_i'$ is not a null vector, but $\sum_i m_i\mathbf r_i' = 0$ is null vector, because by the very definition of center of mass. Because COM balances all unit masses along any axis which cross barycenter. For example, assuming uniform mass distribution in a body and circular shape (for building an easy intuition),- take a look at this example:
![enter image description here](https://cdn.statically.io/img/i.sstatic.net/3yv4X.png)
So, you can see that,
$$ m\left ( (\mathbf r_1' + \mathbf r_5') + (\mathbf r_2' + \mathbf r_6') + (\mathbf r_3' + \mathbf r_7') + (\mathbf r_4' + \mathbf r_8') + \ldots + (\mathbf r_{n-1}' + \mathbf r_n') \right) = 0 \tag 1$$
Or in short,
$$ \sum_i m_i \mathbf r_i' = 0 \tag 2$$
is a null vector.
Second term in your given equation, namely:
$$ \sum_i (\mathbf r_i' \times \mathbf p_i') \tag 3$$
(notice that there goes vector cross-product and only then - sum operation)
is an average relative angular momentum of body around it's COM, hence it's not null, because there's no cancelation of vectors like in the previous case, because each point relative angular momentum around COM "amplifies" other points momentum. Due to the fact that all points cross-product vectors are collinear and points to the same direction.
The point of 1.28 equation is that body angular momentum around any pivot point $O$ can be split into COM angular momentum about that reference point $O$ and body's angular momentum about "itself", i.e. about COM :
$$ \mathbf L_{ref} = \mathbf L_{COM \to O} + \mathbf L_{self \to COM} \tag 4$$