To make an animable character for videogames, is it better to make unique 3D models?

I am making some exercices. Now I have a 3D model of a person with, about, 30k quads, and many separated parts. I want to rig and animate it in order to try it in Unreal Engine or Unity. Is it better to have a unique 3D model for this? I ask, because I am not so able to do manual retopology and I can’t merge all the pieces in one, because all have different numbers of verts, so I can’t use Remove doubles between them, nor Bridge Edge Loops.
So what is the best solution to prepare the model to receive an armature, animations and being exported?

Maybe this would be of use.

This is intense…

Take a look at this, I’m not going to promise that it will be a “quick fix” but this can significantly reduce the amount of time needed to do retopology on models. There are some other tricks to the trade but this will get you started.

Thanks now I consult your links/video, but, so, the answer is that it is better to have a single 3D model, right?
If it isn’t a single model, what are the risks?