Your proof is not correct, see my remarks below.
The calculation becomes relatively simple if one uses the scalar product in $\Bbb C[x]$ defined as
$$
\langle f, g \rangle = \sum_k f_k g_k^*
$$
for $f, g \in \Bbb C[x]$ with $f=\sum_{k} f_k x^k$, $g=\sum_{k} g_k x^k$. It is straightforward to verify that this satisfies all requirements of a (complex) scalar product.
Then
$$
\Vert f \Vert_2 = \sqrt{\langle f, f \rangle}
$$
and therefore
$$
\begin{align}
\Vert (x-z)f \Vert_2^2 &= \langle xf - zf, xf- zf \rangle \\
&= \langle xf, xf \rangle - \langle xf, zf \rangle - \langle zf, xf \rangle + \langle zf, zf \rangle \\
&= \Vert xf \Vert_2^2 - z^* \langle xf, f \rangle - z \langle f, xf \rangle
+ |z|^2 \Vert f \Vert_2^2
\end{align}
$$
and
$$
\begin{align}
\Vert (z^*x-1)f \Vert_2^2 &= \langle z^*xf - f, z^*xf- f \rangle \\
&= \langle z^* xf, z^*xf \rangle - \langle z^*xf, f \rangle - \langle f, z^*f \rangle + \langle f, f \rangle \\
&= |z|^2\Vert xf \Vert_2^2 - z^* \langle xf, f \rangle - z \langle f, xf \rangle
+ \Vert f \Vert_2^2 \, .
\end{align}
$$
These expressions are equal because $\Vert xf \Vert_2 = \Vert f_2 \Vert$: the polynomials $f$ and $xf$ have the same nonzero coefficients, only at different positions.
Regarding your calculation: $(x-z)f$ is the polynomial
$$
\sum_{k=0}^n (x-z) f_k x^k = -z f_0 + \sum_{k=1}^n (f_{k-1}-zf_k)x^k + f_n x^{n+1}
$$
so that
$$ \tag{$*$}
\Vert (x-z)f \Vert_2^2 = |z f_0|^2 + \sum_{k=1}^n |f_{k-1}-zf_k|^2 + |f_n|^2
$$
and that is different from what you calculated for $\Vert (x-z)f \Vert_2^2$.
One can work with $(*)$ to verify the identity $\Vert (x-z)f\Vert_2=\Vert(z^*x-1)f\Vert_2$ and that is effectively the same calculation as what I wrote above. Using the scalar product just makes it simpler.