I retain the relationship $\alpha = \pi - \gamma$ as proposed by JeanMarie; allow me to explain the coordinates of $a$ using 1 variable only:
Let slope of $L_1 = \tan \theta_0$. We have
$$
\begin{align}
a&= (L_1 \cos \theta_0, L_1 \sin\theta_0) \\
b&= (L_3 \cos \theta, L_3 \sin\theta)
\end{align}
$$
Finding the slope of $ab$,
$
\tan \alpha = \frac{ L_3 \sin\theta - L_1 \sin\theta_0
}{ L_3 \cos\theta - L_1 \cos\theta_0
}
$ -- (*)
Hence
$
\cos^2 \alpha =
\frac{(L_3 \cos\theta - L_1 \cos\theta_0)^2
}{(L_3 \sin\theta - L_1 \sin\theta_0
)^2+(L_3 \cos\theta - L_1 \cos\theta_0
)^2
}
= \frac{ (L_3 \cos\theta - L_1 \cos\theta_0)^2
}{ L_1^2 + L_3^2 - 2L_1 L_3 \cos(\theta_0-\theta)
}
$
Differentiate both sides of (*),
$
\sec^2 \alpha \text d \alpha = \frac{
L_3^2 - L_1L_3 \cos(\theta_0-\theta)
}{(L_3 \cos\theta - L_1 \cos\theta_0)^2
} \text d \theta
$
$
\frac{\text d \alpha
}{\text d \theta
} =
\frac{
L_3^2 - L_1L_3 \cos(\theta_0-\theta)
}{(L_3 \cos\theta - L_1 \cos\theta_0)^2
} \cos^2 \alpha
= \frac{
L_3^2 - L_1L_3 \cos(\theta_0-\theta)
}{(L_3 \cos\theta - L_1 \cos\theta_0)^2
} \frac{ (L_3 \cos\theta - L_1 \cos\theta_0)^2
}{ L_1^2 + L_3^2 - 2L_1 L_3 \cos(\theta_0-\theta)
}
= \frac{L_3^2 - L_1L_3 \cos(\theta_0-\theta)
}{L_1^2 + L_3^2 - 2L_1 L_3 \cos(\theta_0-\theta)
}
$
So from $\alpha = \pi - \gamma$,
$\frac{\text d \gamma
}{\text d \theta
} =-
\frac{\text d \alpha
}{\text d \theta
}
$
Hence
$$\frac{\text d \gamma
}{\text d \theta
}
= \frac{L_1L_3 \cos(\theta_0-\theta)-L_3^2
}{L_1^2 + L_3^2 - 2L_1 L_3 \cos(\theta_0-\theta)
}
$$
You can see the value as per asked in the question varies with the size of $\theta$, so there is no definite value.
Edit
By cosine law,
$L_1^2 + L_3^2 - 2L_1 L_3 \cos(\theta_0-\theta) = L_2^2$
I am reluctant in putting $L_2$ in the expression as it is yet another variable.
My objective was to show that $\text d \gamma / \text d \theta$ is not a constant.
Introducing $L_2$ might be confusing as it appears to be a constant like $L_1$ or $L_3$, which isn't in reality.