The usual way the solutions of the Einstien Field Equations are introduced is by saying they are (pseudo-) riemannian metrics that satiafy the diff equations for a given EM Tensor. My question is: what about the actual manifold? Are we assuming a certain type of manifold and then looking for a metric on it, or are we looking for the whole (pseudo-)Riemannian manifold? Because the known solutions all seem to have similar manifolds from a topological POV, but why wouldn't there be solutions with crazy topology? On the other side we do change the manifold for some solutions, for example the Schwarzschild solution removes the center, making it not simply connected anymore right?
So what exactly do we looking for when solving the equations? Maybe manifolds with certain characteristics (for example 3d etc)? I'm terribly confused and would like any clarification on this regards