I don't know if this will fully answer your question, but as a suggestion, you may want to try to solve the problem component-wise assuming that the permittivity matrices are isotropic first (i.e. same value on each diagonal).
In other words, assume something like:
$$
\begin{bmatrix}
D_{x} \\
D_{y} \\
D_{z}
\end{bmatrix}
=
\begin{bmatrix}
\epsilon & 0 & 0 \\
0 & \epsilon & 0 \\
0 & 0 & \epsilon \\
\end{bmatrix}
\begin{bmatrix}
E_{x} \\
E_{y} \\
E_{z}
\end{bmatrix}
$$
but keep each component separate in your calculations. Once you have convinced yourself that you get the correct answer (which I believe you already know how to do), go back and make the matrices anisotropic and redo the calculations. Since you have already seen how the process works component-wise in the isotropic case, this may help guide your calculations in the other case.
As another suggestion, you may want to consult a book like Born and Wolf Optics (https://archive.org/details/principlesofopti00born/mode/1up) Section 14, where they discuss electromagnetic solutions in anisotropic crystals. This may be of some use as well.