Recall that the trace of $B^*A$ is the usual dot product if you regard the matrices $A$ and $B$ as $4$-dimensional vectors $\vec a$ and $\vec b$ in $\mathbb{R}^4$, written as column vectors.
To avoid confusion, we use the notation
$$\langle A,B\rangle_{\text{tr}}=\mathrm{trace}(B^*A)$$
and we'll use $\langle \vec a,\vec b\rangle$ for the dot product. Then
$$\langle A,B\rangle_\text{tr}=\langle \vec a,\vec b\rangle.$$
I'll give you an example. Let
$$A=\begin{pmatrix}1&-3\\4&5\end{pmatrix}\text{ and }
B=\begin{pmatrix}-5&3\\44&15\end{pmatrix}.$$
Now the trace of $B^*A$ is the usual dot product of the vectors
$$\vec a=\begin{pmatrix}1\\-3\\4\\5\end{pmatrix}\text{ and }
\vec b=\begin{pmatrix}-5\\3\\44\\15\end{pmatrix},$$
that is,
$$\langle A,B\rangle_\text{tr}=\langle\vec a,\vec b\rangle=
1\cdot(-5)+(-3)\cdot3+4\cdot44+5\cdot15.$$
Furthermore the norm $\|A\|_\text{tr}$ of $A$ is given by
$$\|A\|^2_\text{tr}=\langle A,A\rangle_\text{tr}=\langle\vec a,\vec a\rangle,$$
as usual.
Now to your question.
Notice that $x$, $y$, and $z$ are mutually perpendicular.
For example,
$$\langle x,y\rangle_\text{tr}=\langle\begin{pmatrix}1\\0\\0\\-1\end{pmatrix}, \begin{pmatrix}0\\1\\0\\0\end{pmatrix}\rangle=0.$$
Moreover, $y$ and $z$ have norm $1$ whereas obviously the norm of $x$ is $\sqrt{2}$. So to achieve a orthonormal basis you only have to divide $x$ by its length, which is $\sqrt2$.