I don't agree. One can directly prove that
$$U(x,y)=\frac{1}{2}\left(y\pm \sqrt{4x-y^2} \right)$$
is solution of the ODE $\quad U_y+UU_x=1$
and if $\quad 4x-y^2=0\quad $ then $\quad U=\frac{y}{2}\quad$ as expected.
So, in contradiction to the statement in the wording of the question, the PDE has a solution (above) such that $\quad U=\frac{y}{2}\quad$ when $\quad 4x-y^2=0$ .
HINT :
Solve the PDE thanks to the method of characteristics leading to the general solution on the form of implicit equation :
$$2x-U^2=F(y-U)$$
where $F$ is any differentiable function.
Then determine the function $F$ so that $y=2U$ and $x=\frac{y^2}{4}=U^2$
The function is $F(X)=X^2$
This leads to $2x-U^2=(y-U)^2$ . Solving for U leads to the above particular solution.