GitLab

#1399 Ensight6: the code has been rewritten slightly to avoid recomputing at each time iteration the match between coords and dofs.

#1399 Introduce a new object which will in next commit also be responsible of computing the match Coords / Dof.

#1399 Add a post-processing test for mesh which vertices aren't all used to bear dofs. I had to broaden slightly the interface of Ensight 6 to enable choosing the output directory.

#1399 Rewrite the post-processing method that writes the dof output for each unknown; the bug leading to ticket #1399 is therefore resolved. I will however write a test and improve it a bit: current implementation was more a proof of concept and is wildly inefficient (some computation done for each time iteration - not an issue on my toy meshes but could be more annoying for heart models...).

#1405 Review the Coords::GetIndex() use and replace some by more appropriate GetPositionInCoordsListInMesh(). Comments on those two methods have been overhauled to make the distinction clearer.

#1404 Yaml: check warnings jobs should not define tags - it; s better to be opportunistic and take the first available (these jobs are very fast and don't use cache).

#1404 Try the suggestion from https://blog.sparksuite.com/7-ways-to-speed-up-gitlab-ci-cd-times-29f60aab69f9 to use another Docker driver to speed-up CI.

#1398 It's actually not possible to use a define inside a macro (see https://stackoverflow.com/questions/7246512/ifdef-inside-a-macro). I therefore had to duplicate the macro for the clang and not clang cases... (using a middleman macro didn't work as the warnings about macro recursion popped up - the very same for which I had to disambiguate clang and gcc in the first place in the macro...)