GitLab

**Note: This only affects the skeleton not the boundary term!**
Upwinding can now be turned off (`none`)
or set to full upwinding (`fullUpwind`).
Removed 'weights' option.

New options are OOP (no penalty term) and IIP (no symmetry term)
Also adapted error estimator (for OOP scheme)
Moved error estimator into Operator namespace
Changed initialization orders in both operators to suppress warnings

Time parameter has to be the conclusion time of the last time step.

[ci skip] because this branch has the old testing scheme

The time variable was unused, which caused the builds to fail.

Unquaring must occur outside the error indicator because summation
occurs over multiple intersections.

FlowBoundary and time now need to be passed to the error indicator.

Error indicator now does not contain squares anymore.
Removed VTK output of error estimators.
Removed all unused variables inside the estimator.

changed newest adaptation scheme to skeleton term proposed by diPietro and Ern. writing this and old error residuals into extra vtk files

tests are allowed to fail until they are adapted.
'deploy' stage only contains building the docs.

see #5

fixes #31

Revert "rolled back to requirements.txt files for supporting reinstallation of the dorie python modules"

This reverts commit b314fbc6.

* updated requirement list
* added optional package list
* reworked manual installation
* warning not to call 'make install' #16
* notes on parallel setup requirements #24

removed ode tests with adaptation for now. increased adapt threshold for ref tests to account for new estimator

changed evaluation positions of saturation and condfactor at intersections for upwinding condition, fixing #26

corrected numeric quadrature in error estimator. added face diameter weighting for error indicator value. this leads to over-refinement in old settings

adaptivity GFS uses the same FEM type as problem GFS

See merge request !12