* 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

* in dark mode contents of latex generated formulas would be basically invisible

* this should espectially work well on vector machines
* it is enabled by compiling with "-DYAC_NEC_EXPERIMENTAL"

* routine yac_get_cell_bounding_circle uses specialised routines for most common cell type
* should improve performance

* originally this routine also computed the very expensive sine of the angle...

Moritz Hanke authored 1 month ago and Nils-Arne Dreier committed 1 month ago

Co-authored-by: Nils-Arne Dreier <dreier@dkrz.de>

* returns interpolation information for raw data exchange in a CSR format