move transform calls to models, add properties to adjust behavior

[ArcRiley]

currently Scene calls each Body._render, which performs it's translation and then calls it's ._model._render

This poses a hurdle for models like Billboard, which have to correct this translation. Further, the above sequence uses a function call (to _body) which does very little in itself.

Instead, Model._render should be called directly from the Scene. Special translation modes, such as billboarding, should be a property which many models can set independent of what they actually render.