Support multiple visuals per link in URDF model

[swhart115]

The urdf parser being used only supports one "visual" element, per "link" element, but URDF supports an arbitrary number. This affects path visualization, and leads to gaps in the display (see the Atlas forearms as an example: there are 5 or 6 primitive shapes associated with this link, but only the first can be displayed now).