* average with barycentric coordinates weightening could case an error
  for regular source cells, if one of the edges is on a pole

* HCSBB could cause an issue for regular cells that has two of its vertices at a pole. The triangulation would cae on of the two triangle to be a single line, which was not correctly handled.

now takes areas differences between edges following circles of constant latitude and edges following great circles into account during barycenter computation

* some reordering of operations
* improved documentation