1) Since $X\geq 0$ and by the definition of conditional expectation,
$$E(X|G)=\int^{\infty}_0 x f(x|G) \mathrm{dx}$$
We can see that $x f(x|G)\geq 0$ for all $x\in [0,\infty)$, since $x\geq 0$ and $f(x|G)\geq 0$ by definition. Since the function is nonnegative on the whole interval it's being integrated over, we can deduce that the integral is nonnegative as well.
2) A constant doesn't take on any values with probabilities, since it's not "random", hence why we call random variables random - they take on values with certain probabilities. So the expected value of something that doesn't change is itself.
3) In the expression $E(X|Y)$, we integrate over all possible values of $X$, since $Y$ is observed, i.e., we know what value $Y$ is. Now, using this fact, you can deduce that $g(Y)=a$, for some $a\in \mathbb{R}$. Furthermore, $E$ is linear, meaning $$E(aX+bY)=aE(X)+bE(Y)$$
So $E(G(Y)X|Y)=E(aX|Y)=aE(X|Y)=g(Y)E(X|Y)$
If you haven't already, take a look at this and this question.