Recall that
\begin{aligned}
\mathcal{J}_{k-\frac{1}{2}}((\mu_1-\mu_2)z)&=\frac{\Gamma(2k)}{2^{2k-1}\Gamma(k)^2}\int_{\mathbb{R}}
e^{(\mu_1-\mu_2)y}(1-\frac{y^2}{z^2})^{k-1}\chi_{[-1,1]}(\frac{y}{z})z^{-1}dy,\\
\mathcal{J}_{k-\frac{1}{2}}'((\mu_1-\mu_2)z)&=\frac{\Gamma(2k)}{2^{2k-1}\Gamma(k)^2}\int_{\mathbb{R}}
e^{(\mu_1-\mu_2)y}(1-\frac{y^2}{z^2})^{k-1}\frac{y}{z^2}\chi_{[-1,1]}(\frac{y}{z})dy.
\end{aligned}
Inserting these into (1) and making use the change of variables:
x=\frac{\nu_1+\nu_2}{2},\quad z=\frac{\nu_1-\nu_2}{2},
with Fubuni’s Theorem, we obtain
\begin{aligned}
E_{k}(\mu,\lambda)= \int_{\mathbb{R} }\int_{\mathbb{R} }
e^{(\mu_1+\mu_2-2\mu_3)x+ (\mu_1-\mu_2)y} F_k(x,y,\lambda)dxdy
\end{aligned}
where
\begin{aligned}
&&F_k(x,y,\lambda)=\\&&
\frac{\Gamma(2k)\Gamma(3k)}{2^{2k-2}\Gamma(k)^5V(\lambda)^{2k}}\int_{\mathbb{R} }\Big( 3z^2(\lambda_1-\lambda_2)-6y(x^2-z^2+\lambda_3x+\lambda_1\lambda_2)\Big)
\left(\frac{(\lambda_3-x)^2-z^2}{z^2}\right)^{k}\\&&
\Big((z^2-y^2)(\lambda_1-x)^2-z^2)(z^2-(\lambda_2-x)^2)\Big)^{k-1}\\&&\qquad\qquad\qquad\qquad\qquad\qquad\qquad
\chi_{[-1,1]} (\frac{y}{z}) \chi_{[\lambda_1,\lambda_2]}(x+z)\chi_{[\lambda_3,\lambda_2]}(x-z)dz\\
=&&\frac{\Gamma(2k)\Gamma(3k)}{2^{2k-2}\Gamma(k)^5V(\lambda)^{2k}}\int_{\max(|y|,|x-\lambda_2|)}^{\min(x-\lambda_3, \lambda_1-x)}
\Big( 3z^2(2y+\lambda_1-\lambda_2)-6y(x-\lambda_1)(x-\lambda_2)\Big)\\&&
\qquad\qquad\qquad\qquad\qquad\left(\frac{(\lambda_3-x)^2-z^2}{z^2}\right)^{k}
\Big((z^2-y^2)((\lambda_1-x)^2-z^2)(z^2-(\lambda_2-x)^2)\Big)^{k-1}dz
\end{aligned}
where we used the fact that
\begin{aligned}
\chi_{[-1,1]} \left(\frac{y}{z}\right) \chi_{[\lambda_1,\lambda_2]}(x+z)\chi_{[\lambda_3,\lambda_2]}(x-z)=
\chi_{\max(|y|,|x-\lambda_2|)\leq z\leq \min(x-\lambda_3, \lambda_1-x)}.
\end{aligned}
Now, the change of variables
x=\frac{\nu_1+\nu_2}{2},\quad y=\frac{\nu_1-\nu_2}{2},
gives
\begin{aligned}
E_{k}(\mu,\lambda)= \int_{\mathbb{R} }\int_{\mathbb{R} }
e^{\langle\mu,\nu \rangle} F_k\left(\frac{\nu_1+\nu_2}{2},\frac{\nu_1-\nu_2}{2},\lambda\right)d\nu_1\nu_2.
\end{aligned}
To achieve the proof we use that
\left\{\nu\in \mathbb{V}; \quad \max \left(\frac{|\nu_1-\nu_2|}{2},\left| \frac{\nu_1+\nu_2}{2}-\lambda_2\right|\right)\leq \min
\Big(\frac{\nu_1+\nu_2}{2}-\lambda_3,\lambda_1-\frac{\nu_1+\nu_2}{2}\Big)\right\}=co(\lambda).