Feature/fix boulernm efficiency

[Algiane]

This PR rewrites the setVertexNmTag (function that analysis the type of feature edges arriving through a non-required non-manifold point and set the suitable tag at point):

• initial function was travelling the tetra vertices, selecting the edges of the ball of the vertex incident to the vertex and counting the feature edges (edges where hashed in a new hash table for each vertex to be uniquely added);
• now we travel tetra edges and increment edge vertices counters depending on the type of the feature edges. Edges are hashed to be uniquely added. Then we travel the mesh points and set the suitable tags depending on the computed counters.

This PR removes the boulernm function.

[prj-]

@Algiane, this PR breaks ParMmg.

Undefined symbols for architecture arm64:
  "_MMG5_boulernm", referenced from:
      _PMMG_update_singul in analys_pmmg.c.o

[Algiane]

Thanks for this report. I will try to fix this as soon as possible (I am in business travel for now).

[Algiane]

Hi,
Thanks for your patience. I have merged a temporary fix in Mmg (see issue #97 ).
I hope that it will work while we wait for a real solution to the problem.
Best

[prj-]

Yes, it does work, thanks.

[prj-]

Sorry, I spoke too soon. Though now ParMmg builds successfully, it fails a very basic FreeFEM test with the following error.

Assertion failed: (isfinite(dd)), function PMMG_hashNorver_normals, file analys_pmmg.c, line 990.

[Algiane]

Hi,

Sorry for that. I will try to find this issue before the summer hollidays.

Best

[prj-]

No problem, in the meantime, I'm sticking to commit 6824158449e57b46253b505d968b81f15fafc6a0 of Mmg. Let me know if you want me to try something that you cooked up, there is no hurry at all, I'll still be here after the summer holidays :)