This is not a complete solution, but just presenting a few ideas.
Let's work on $\rho$ first:
\begin{align*}
\rho(u)&=\frac{\sin(u-\lambda)}{\sin(\lambda)}\\
&=\frac{\sin(u)\cos(\lambda)-\cos(u)\sin(\lambda)}{\sin(\lambda)}\\
&=\sin(u)-\cos(u),
\end{align*}
since $\lambda=\pi/4$ and $\sin(\pi/4)=\cos(\pi/4).$
Next, we look at what happens when we have $\rho(u)\cdot\rho(-u).$ We have
\begin{align*}
\rho(u)\cdot\rho(-u)&=[\sin(u)-\cos(u)][-\sin(u)-\cos(u)] \\
&=-\sin^2(u)-\sin(u)\cos(u)+\sin(u)\cos(u)+\cos^2(u)\\
&=\cos^2(u)-\sin^2(u)\\
&=\cos(2u).
\end{align*}
From this, we gather that
\begin{align*}f(u)\,f(u+\lambda)&=\prod_{i=1}^L\rho(u-u_i)\,\rho(u_i-u)+\prod_{i=1}^L\rho(u+\lambda-u_i)\,\rho(u_i-u-\lambda)\\
&=\prod_{i=1}^L\cos(2(u_i-u))+\prod_{i=1}^L\cos(2(u_i-u-\lambda)) \\
&=\prod_{i=1}^L\cos(2(u_i-u))+\prod_{i=1}^L\cos(2(u_i-u)-\pi/2),\;\text{or} \\
f(u)\,f(u+\pi/4)&=\prod_{i=1}^L\cos(2(u_i-u))+\prod_{i=1}^L\sin(2(u_i-u)).
\end{align*}
From this, we can at least see that if $u=u_i$ for any $1\le i\le L,$ then the second product drops out and you get
$$f(u_i)\,f(u_i+\pi/4)=\prod_{\begin{array}{c}j=1\\ j\not=i\end{array}}^L\cos(2(u_j-u)).$$
Conversely, if $u_i-u=(2k+1)\pi/4$ for some $k\in\mathbb{Z},$ then the first product series drops out and you only get the second one.
Finally, it's always worthwhile plugging in $u=0$ to arrive at
$$f(0)\,f(\pi/4)=\prod_{i=1}^L\cos(2u_i)+\prod_{i=1}^L\sin(2u_i).$$
The RHS of
$$f(u)\,f(u+\pi/4)=\prod_{i=1}^L\cos(2(u_i-u))+\prod_{i=1}^L\sin(2(u_i-u))$$
is $\pi$ periodic, and hence the LHS is as well.
This is however not efficient when $L$ becomes large and I was hoping to find something like a closed form, non-linear integral equation or anything else which is fast also for large $L$.
– TheoPhysicae Mar 28 '19 at 12:17