- clean up GModel API (xxByTag -> getXx, numXx -> getNumXxes, ...)

allow multiple optimization passes

move random seed init in GenerateMesh

start reducing header pollution (removed Gmsh.h + cleaned up GModel & co)

fix infinite loop

moving to new post-processing database

                    *** WARNING ***

do not upgrade if you need plugins or adaptive views: these have not
been ported (yet) to the new database

                    *** WARNING ***

lots of improvements to boundary layer generation: we can now constuct
and mesh models with entities connected to boundary layer points, edges
and faces

- removed static GMODEL (can now have multiple models)
- added infrastructure for new post-processor module

add big warning for Delaunay algo

- prepare group of nodes in UNV (for Guillaume)
- prepare delaunay for connected volumes
- test 2nd patch from Michel for phase1() extrusion

bug fixes in new high order element code

new boundary layer mesh generator

refactored smooth value/smooth normal code into single SmoothData.* file

- print num vert+elm after each generation
- reork tutorial a bit

fix logic for 3d delaunay

- reimplemented subdivision of extruded meshes
- fix indent + unused var + gcc warnings all over the place

free vol mesh befire surf mesh

cleanup msg

change the logic in GenerateMesh():

- never reload the model file--we don't need this trick anymore!

- always only generate the mesh we are asked to generate (pressing
  "1D" should *always* mesh the edges; pressing "2D" should only mesh
  the surfaces, unless there is no 1D mesh). This makes the behavour
  much more predictable when changing mesh attributes interactively.

small fixes for mesh opti

cleanup

bidir db: phase 3 -- move all the old geo stuff into Geo/

bidi db: phase 2 -- continued

bidir db: phase 2 -- removed all (well, most of) the old data structures and
algorithms

force orientation of mesh to match the orientation of the geometry

added full support for incomplete second order elements (8-node quad,
20-node hexa, 15-node prism and 13-node pyramid):

- second order generation algorithm can now be told to generate either
  complete or incomplete elements

- MSH I/O routines can deal with all elements

- UNV I/O routines now support all the elements in the I-deas spec (I did
  my best to follow the spec for the ordering of the nodes, but I might have
  made some mistakes: PLEASE TEST IF YOU HAVE ACCESS TO A CODE OTHER THAN
  GMSH THAT CAN READ/WRITE UNV FILES)