According to Halmos (Lectures on Boolean Algebras, p. 92), the completion of Boolean algebra $A$ is a complete Boolean algebra $B$ together with a monomorphism $h$ from $A$ into $B$ such that
(1) $h$ preserves all suprema.
(2) The complete Boolean algebra generated by $h(A)$ in $B$ is $B$ itself.
Moreover, a completion $(B,h)$ is minimal if, corresponding to every completion $(C,k)$, there exists a complete monomorphism $f$ from $B$ into $C$ such that $f\circ h=k$
Question 1: If $h(A)$ generates B, then, by definition, B is the smallest complete Boolean algebra containing $h(A)$. Why is this fact not a sufficient condition to define minimality?
Question 2: Is there another way to define minimality (than the one provided above)?
You should cite Halmos properly. Halmos does not define the completion of a Boolean algebra", but a completion of a Boolean algebra. Then he defines a minimal completion using a universal property and insists that
Question 1. The fact that the complete Boolean algebra generated by $h(A)$ in $B$ is $B$ itself does not suffice to insure that $B$ satisfies the universal property.
Question 2. You could use Halmos' Theorem 11 from the same book as a definition, but it is not really simpler. In this case, you would define the completion of a Boolean algebra $A$ as the Boolean algebra of regular open sets of the Stone dual of $A$. But you could also use another result, known as the Glivenko–Stone theorem: