I think it is important to realize that reality is special. There is some structure to it, whether we can figure out what it is or not. And there is even radically different good approximations at different scales. At the human scale, objects in space seem to obey euclidean geometry. What does this mean? Well, for one thing, rigid motions (combinations of translations and rotations) preserve lengths and angles! Have you ever wondered if a solid object can be moved in some manner that its shape is changed? It just does not. Under any rigid motion $f$ in euclidean geometry, $|f(P)-f(Q)| = |P-Q|$ for every points $P,Q$, and $∠PQR = ∠f(P)f(Q)f(R)$ for every points $P,Q,R$. The fact that objects at the human scale seem to obey this tells us that euclidean geometry is a good approximation of the geometry of reality at the human scale (not too big or small). In fact, not just euclidean geometry by specifically the euclidean space $ℝ^3$ seems to be a good approximation for human-scale geometry of reality. And $ℝ^3$ is a vector space and obeys vector addition in the sense that you noted.
In other words, we would not be so interested in $ℝ^3$ and vector spaces in general if reality did not in the first place have structure that looks like $ℝ^3$ at a scale we can observe! Similarly, you could ask why people came up with the axiomatization of the naturals as a discrete ordered semi-ring plus induction, and the answer is again the same! Namely, counting seems to obey these properties!
As for the addition of forces specifically, it stems from the relation between force and acceleration in newtonian mechanics (which is at the human-scale). We have that a force $F$ on a mass $m$ induces acceleration $a$ given by $F = m·a$, and $a$ is the second derivative of position $x$. Now a position is a vector in $ℝ^3$ (according to our good approximation), so acceleration is also a vector in $ℝ^3$, and adding two forces is observed to induce the sum of the corresponding accelerations, and hence two forces applied on the same mass induce exactly the same acceleration as the vector sum of those forces applied on that mass. Take a while to think through this; it is indeed not merely an issue of discovery, but there is some underlying mathematical constraint given the observed relation between forces and positions via acceleration.