Since we know our pendulum is not getting any longer, we know that the net force in the radial direction must be zero. And so the net force acting on the pendulum is in the direction of $\hat{\theta}$ as you mentioned. This force is obviously going to be $( -mg \sin{\theta} )\hat{\theta} $ , which of course gives us:
$$m\vec{a}=( -mg \sin{\theta} )\hat{\theta} $$
And since $a$ is in th rotational direction, we can say $\vec{a}=(l \frac{\mathrm{d}^2}{\mathrm{d} t^2}\theta) \hat{\theta}$
Plugging that back in we obtain:
$$\frac{\mathrm{d}^2}{\mathrm{d} t^2}\theta = -\frac{g}{l}\sin{\theta}$$
This differential equation tells you everything about the pendulum, and you can use a small angle approximation to obtain a Harmonic Oscillator, but if you are interested in centripetal force it will be a little more subtle since we are now dealing with non-uniform circular motion, and so the velocity vector is not so nice to deal with. This is because in uniform circular motion $v$ is just a number, however in the case of a pendulum its a time varying vector.
But to understand why in general $a_c=\frac{v^2}{r} \rightarrow F_c=ma_c=\frac{mv^2}{r}$ consider the following:
Triangles ABC and PQR are similar and therefore:
$$\frac{\Delta v}{v}=\frac{\Delta s}{r}\rightarrow a:=\frac{\Delta v}{\Delta t}=\frac{v}{r}\frac{\Delta s}{\Delta t}= \frac{v^2}{r}$$
And so we obtain $a_c=\frac{v^2}{r}$