Re the units of $A$: $A$ has dimensions of $L^{-3/2}$. The modulus of $\Psi^2(\vec{x},t)$ is a probability density so it has units of probability per unit volume, i.e. it's dimension is $L^{-3}$, so the dimensions of $\Psi$ and therefore $A$ must be $L^{-3/2}$. I don't think there is an SI unit corresponding to $\sqrt{P/m^3}$, but then it's hard to see when you would need such a unit.
Re the other question: the function:
$$\Psi(\vec x, t)=Ae^{i(\phi+\vec k\vec x-\omega t)}$$
is indeed a solution to the Schrodinger equation for a free particle, but you need to be careful about it's physical meaning. As mentioned above, the modulus of $\Psi^2$ gives you the probablity of finding the particle per unit volume, but $\Psi$ itself doesn't have a physical interpretation. You're quite correct that the equation can also be written:
$$\Psi(\vec x, t)=Acos(\phi+\vec k\vec x-\omega t) + Aisin(\phi+\vec k\vec x-\omega t)$$
but you can't attach any physical meaning to the two terms on the right.
P = probability
Re the comment: I suppose the SI unit would be the metre$^{3/2}$, but the value of the wavefunction has no physical significance so no such unit has ever been defined.
This is why the value has no physical significance: suppose $\phi$ is zero, then what is $\Psi(0, 0)$. The answer is of course $A$. But the phase $\phi$ has no absolute meaning, it's a phase relative to some reference point that can be freely selected. So suppose I use a reference point that is $\pi/2$ different to yours then I'd calculate $\Psi(0, 0)$ to be $iA$ not zero.
So you think $\Psi(0, 0) = A$ while I think $\Psi(0, 0) = iA$. Does this make any difference in the real world. Well the probability deniity is $\Psi^*\Psi$ so you calculate the probablility density to be $A^2$, and I calculate the probability density to be, erm, $A^2$ and we get the same answer even though we have different values for $\Psi$.