In the case of a black hole, trying to draw a line from the singularity to the edge of the event horizon is pointless in regular spherical coordinates (though, theoretically, you could try in Kruskal-Szekeres coordinates). We define the event horizon radius not as a quantity on its own but the circumference of the event horizon sphere divided by $2\pi$.
For less-extreme objects you can still speak of radius. A star for example does curve space around it that would distort measurements, but the measurements are made locally and by an observer embedded in the spacetime. You, as a human in spacetime, might measure a distorted diameter for the star, but from the perspective of someone off the manifold, you can still pick two antipodal coordinate points on the star’s surface and take their Euclidean distance to be the diameter of the star, even though that distance might not exactly correspond to how a human would measure it.
I think that, physically, it is more meaningful to describe something as it can be perceived at infinity (i.e. at rest relative to it, very far away so we are “outside” the influence of its gravitational effects and what have ye). Close to it, one might experience the geometry differently (see time flow being different closer to/farther from a black hole). In most cases, though, we are far away from the things we’re talking about, if we’re talking about stars or quasars or black holes or alien warp drive spaceships. This perception from infinity might be distorted, but it’s the best thing you can get from a measuring-the-thing-from-a-particular-point perspective.