First, here is a general lemma:
Lemma: Let $I,J\subseteq R$ be ideals and suppose $I\cap J$ and $I+J$ are finitely generated. Then $I$ and $J$ are finitely generated.
Proof: Let $x_1,\dots, x_n$ generate $I\cap J$, and let $y_1+z_1,\dots,y_m+z_m$ generate $I+J$, for $y_i\in I$, $z_i\in J$. If $w\in I$, then $w\in I+J$, so we can write $w=\sum c_i(y_i+z_i)$. Note that $\sum c_iz_i=w-\sum c_i y_i\in I\cap J$, so we can write $\sum c_iz_i=\sum d_j x_j$. We have now written $w=\sum c_iy_i+\sum d_j x_j$. Since the $y_i$ and $x_j$ are all in $I$ and $w\in I$ was arbitrary, this means $\{x_1,\dots,x_n,y_1,\dots,y_m\}$ is a finite set of generators for $I$. By swapping the roles of $I$ and $J$, we see $J$ is also finitely generated.
To solve your problem, we will apply the Lemma to $I=\mathfrak{a}$ and $J=(a)$. You have already noted that $\mathfrak{a}+(a)$ is finitely generated. To show that $\mathfrak{a}\cap (a)$ is finitely generated, first consider the ideal $K=\{c\in R:ac\in\mathfrak{a}\}$. Clearly $\mathfrak{a}\subseteq K$, and by hypothesis $b\in K$ as well, so $K$ is strictly larger than $\mathfrak{a}$ and must be finitely generated. But $aK=\mathfrak{a}\cap(a)$, so this implies $\mathfrak{a}\cap(a)$ is also finitely generated. By the Lemma, we get that $\mathfrak{a}$ is finitely generated, which is a contradiction.