Compute sum_k M[k] @ V[k], the sum of the p matrix products, and return the (n, 1) result:

• M holds p matrices of shape (p, n, n)
• V holds p vectors of shape (p, n, 1)
• use a single vectorized call, not a Python loop