Add affine_map and affine_map_inverse to finite_element

GNW: Do we need affine in the name? It would be nice to have something that would later generalise to non-affine maps.

AL: If the mapping is different from affine, we would need more data than what is currently stored in ufc::cell. Since it is not clear at this point how that should be stored, I suggest we add affine_map now and either replace it with a more general function later, or add something like higher_order_map alongside affine_map later.

AL: I'm renaming these as follows:

  map_from_reference_cell
  map_to_reference_cell

Other functions in the interface that do something start with what they do: tabulate_, evaluate_, interpolate_, so the functions should start with map_.