Reading about gravitational lensing here and here, its seems that it can be explained well by assuming a refractive index created by a gravitional potential $$n(\vec{x}) = 1+2\frac{\phi(\vec{x})}{c^2}$$ where $\phi$ is the gravitational potential (negative values, $\phi\to-0$ for $|\vec{x}|\to\infty$).
Looking at this answer or this, this again is the same as the $g_{00}$ term of the metric tensor.
Due to symmetries, the metric tensor has 10 free variables. Now I wonder: if gravitational lensing could indeed be described by just one number (per space point $\vec{x}$), i.e. a scalar field, how come the metric tensor has 10 free variables (per spacetime point)? What are they "good for"? In which situations does the refractive index not tell the whole story?