Well I am working on my own game and I was wondering if it is a good move to make the humanoid in separate parts? Let's say I create the head and paint it separately, I create and paint separately the legs, hands, accessories...etc and then combine the UVs and make an atlas with those textures. I find it easier for me at least and I do not get lost in a big mesh.
Generally speaking it's not how 3D game artists are building their models. Separating the head and the limbs from the body might make things easier in the beginning of the modeling process but later on it will be harder to produce a good result.
Separate limbs always result into seams. You can hide them under garments but you cannot get rid of them by painting over them.
As for the aspect of rigging and animating them you might want to take into account that most of the rigging tools (for Blender and otherwise) are designed to work with single meshes. At one point you will have to weight-paint your mesh to get a good result. Having two separate meshes even in the same model leads to gaps and deformities which cannot be fixed. It's just the way inverse kinematics work.
About that "getting lost in a big mesh" part - you can always hide parts of your mesh in edit mode. Select the part you want to hide, then hit H to hide it and Alt+H to unhide it.