I'm trying to calculate the inertia tensor for a circular arc, as shown in this image:
It starts at the x-axis and lies entirely in the XY plane. It has a mass per unit length of $\rho$ and spans an angle of $\theta$ (which could be anything up to $\pi$).
I've found many resources which give the inertia tensor for a full hoop, or for a segment, but not this case. I think the off-diagonal terms are zero, and that $I_{zz}$ is just a fraction of the polar inertia for a full hoop, so: $$I_{zz} = \frac{\theta}{2\pi}\left(\rho 2 \pi r\right) r^2 = \rho \theta r^3$$ ... but I can't work out $I_{xx}$ or $I_{yy}$, or even if they're the same as each other.

The reference system is not centered at the barycenter of the arc, and it is not oriented along the symmetry axes.
Therefore, you did right for $I_{zz}$, but for the rest you do have off diagonal elements, and you cannot do else than calcutating all the nine (well, six) components.