This question is a follow-on of a question asked today that has been erased by its author an hour after being asked, for unknown reasons. I found it interesting ; I have a solution, mainly based on analytic geometry. I would appreciate to see other solutions (using synthetic geometry for example), hopefuly less computational.
Let me reformulate it in a slightly different way compared to the original question :
Given : a fixed line with two points $B$ and $C$ on it. Let us construct $A$ such that $ABC$ is equilateral. Let us take an arbitrary point $F$ on line $BC$ outside line segment $[BC]$. Let us denote by $D$, resp. $E$, the center of the incircle of triangle $FAC$, resp. the center of the excircle of triangle $FAB$ opposed to $F$.
Prove that
$$BE+CD=BC\tag{1}.$$
Solution :
Let us define $A'$ as being the symmetrical point of $A$ with respect to line $BC$. $A',C,D$ are aligned. Indeed, slope(A'C)=slope(C'D) for the following reason : as $\angle ACF = 180°-60°=120°$, we have $\angle DCF=\angle DCA=\angle ACB = 60°$.
For a similar reason, $A',B,E$ are aligned.
Let us call $D',E'$ the projections of $D,E$ resp. on line $AB$ ; (1) amounts to say that :
$$BE'+CD'=\tfrac12 BC\tag{2}.$$
Let us take coordinate axes such that :
$$B(-1,0),C(1,0),A(0,\sqrt{3}),F(a,0) \ \text{with} \ a>1$$
as one can see on the figure.
Let $2 \alpha = \angle AFB$. We have
$$\tan(2 \alpha)=\frac{\sqrt{3}}{a}$$
Using relationship $$\tan(2 \alpha)=\frac{2 t}{1-t^2} \ with \ t:=\tan(\alpha)\tag{3}$$
we get :
$$t:=\tan(\alpha)=\frac{\sqrt{3+a^2}-a}{\sqrt{3}}$$
(but in fact, we will not use this relationship).
Point $D$ being at the intersection of :
$$\begin{cases}\text{Line FD : } & y=-t(x-a)\\ \text{Line A'C : } & y=\sqrt{3}(x-1)\end{cases}$$
its abscissa, which is the same as the abscissa of $D'$, is
$$x_{D'}=\frac{at+\sqrt{3}}{t+\sqrt{3}}$$
For a similar reason :
$$x_{E'}=\frac{at+\sqrt{3}}{t-\sqrt{3}}$$
(2) will be established if we show that :
$$\left(-1-\frac{at+\sqrt{3}}{t-\sqrt{3}}\right)+\left(\frac{at+\sqrt{3}}{t+\sqrt{3}}-1\right)=1\tag{4}$$
With some simplifications, the LHS of (4) becomes :
$$(at+\sqrt{3})\frac{-2 \sqrt{3}}{t^2-3}-2\tag{5}$$
But, using (3),
$$t^2-1=\frac{-2ta}{\sqrt{3}}\tag{6}$$
Plugging (6) into (5) gives this LHS equal to $1$ as awaited.
Edit : I hadn't understood at first the track followed by the asker of the initial question, based on the sine law ; it can be completed by checking that for all $\theta$ :
$$\frac{\sin(\theta)}{\sin(2\pi/3-\theta)}+\frac{\sin(\pi/3-\theta)}{\sin(\pi/3+\theta)}=1.$$



