$P \in \mathbb{R}^{NxN}$ is a Projection Matrix, so $P^2=P$.
So the first task was to calculate all possible Eigenvalues.
I got $\lambda_1=1$ and $\lambda_2=0$.
Now the task is: Find a Projection $P: \mathbb{R}^2 \Rightarrow \mathbb{R}^2$, that has the calculated Eigenvalues
So in the solutions it's written:
$P: \mathbb{R}^2 \Rightarrow \mathbb{R}^2$
$P(\begin{matrix} v_1\\ v_2 \end{matrix})=(\begin{matrix}v_1\\0\end{matrix})$
$P(\begin{matrix} a\\ 0 \end{matrix})=(\begin{matrix}a\\0\end{matrix})= 1*(\begin{matrix}a\\0\end{matrix})$
$P(\begin{matrix} 0\\ b \end{matrix})=(\begin{matrix}0\\0\end{matrix})= 0*(\begin{matrix}0\\b\end{matrix})$
How did they come to this?
Is it because $P* x = \lambda * x$?
As you say, projections can only have eigenvalues $0$ or $1$. Thus the task here is simply to present an example which has both, right?
Guess-and-check would be a natural first thing to try. If I ask you to think of a projection $P:\mathbb R^2 \rightarrow \mathbb R ^2$, chances are the first one you'll come up with would be $$ P\begin{bmatrix}x\\y\end{bmatrix} = \begin{bmatrix}x\\0\end{bmatrix} $$ Computing the eigenvalues of $P$ shows that it does, in fact, have both $0$ and $1$ as eigenvalues.