In most general relativity texts, the singularity is treated as a point removed from the manifold, to avoid having to deal with the infinite curvature of the Ricci scalar.
But in the case of a more realistic scenario (let's say a spherically symmetric collapse, for instance), does this idea really hold up? We start with a, let's say, (partial) Cauchy surface, with topology $\Bbb R^3$. The matter collapses, leading to a black hole. I'm not quite sure of the topology of a non-maximally extended black hole, especially one from stellar collapse (in particular I'm not sure if the singularity matters all that much since I have a suspiscion it might not affect the topology, as it may be at timelike infinity). If the topology of the spacelike hypersurface is of the form $\Bbb R^3 \setminus \{0\}$, does that not violate theorems on topology change? As I'm pretty sure it will not violate conditions of that theorem on time orientability or closed timelike curves, does the metric have a degenerate point?
I know that there are also ways to treat singularities without removing the point (ie with generalized functions), and that the problem probably doesn't happen in quantum gravity, but is it a problem when it comes to classical general relativity?