As far as I know “draw calls” are the deciding factor. The engine normally loads a texture and paints it everywhere where it is needed in one draw call. So having one larger texture is a benefit.
It might be a disadvantage when you combine textures which have nothing to do with each other, for example combining a character texture with a texture of a house because you are wasting GPU memory when not both objects are in the same scene. This is always the case when you only use a part of the texture.
Furthermore use material instances as often as possible instead of materials because they are also more performant.
So short answer. It depends 