First of all, in order to have an idea of the behavior of this differential system, I have obtained, using a Matlab script, the graphics below. One can see in particular on this figure the $2 \pi$ periodicity in $x$ and $y$ of the solutions.
Point $x=\pi/2,y=-\pi/2$ is not a solution (the arrows are not zero at this point), but there is a rather close solution as we will see it.
Your error comes from the fact that, having a system of 2 equations with two unknowns
$$\begin{cases}\sin(x-y)&=&\sin(x)\\ \sin(y-x)&=&\sin(y)\end{cases} \ \ \iff$$
$$\tag{1}\begin{cases}\sin(x-y)&=&\sin(x)& \ \ \ (a)\\ \sin(x-y)&=&\sin(-y)& \ \ \ (b)\end{cases}$$
you have reduced it to a single equation $\sin(x)=-\sin(y)$, and thus obtained an erroneous conclusion.
Here is a way to solve system (1). As
$$\sin(u)=\sin(v) \ \ \ \iff \ \ \ u=v+2k\pi \ \text{or} \ u=\pi-v+2k\pi $$
system (1) is equivalent to four "coupled" solutions:
$$\begin{cases} x-y&\equiv&x \ &\text{and}& \ x-y&\equiv&-y \ &\implies& x&=&2k\pi,& y&=&2\ell\pi,\\
x-y&\equiv&\pi-x \ &\text{and}& \ x-y&\equiv&-y \ &\implies& x&=&2k\pi,& y&=&\pi+2\ell\pi,\\
x-y&\equiv&x \ &\text{and}& \ x-y&\equiv&\pi+y \ &\implies& x&=&\pi+2k\pi,& y&=&2\ell\pi,\\
x-y&\equiv&\pi-x \ &\text{and}& \ x-y&\equiv&\pi+y \ &\implies& &&&&&(\text{see below.})
\end{cases}$$
for $k \in \mathbb{Z}$
Detailed solution of the fourth case:
If $x-y=\pi-x+2K\pi$ and $x-y=\pi+y+2L\pi$, we can infer that $y=-x +2M\pi$. Then
$2x=\pi-x+2K\pi$. Thus
$$x=\frac{\pi}{3}+k\frac{2\pi}{3}, y=-\frac{\pi}{3}-k\frac{2\pi}{3} \ \ \text{with the same} \ k.$$
for $k \in \mathbb{Z}$, as can be verified on the graphics.
Remark: on this figure, some integral solutions have been displayed.
