Old GIMLi apps removal

[halbmy]

There are a couple of old C++ apps that we should get rid of. Some of them (bms2vtk, invlinearmat, polyfit) are too simple and available as Python apps, examples, docstrings or code snippets, so they could be removed straight away. For some (ttinv, dc1dinv, fdem1dinv, fdem1dsmooth, mrs1dblock, mrs1dsmooth) there are existing method managers that can also be called as apps, so there is also no need to keep them. For some others (mt1dinv, mt1dsmooth) there are no managers neither examples yet. (Independent of this we should provide managers for all provided geophysical methods) We could keep all of them as the apps are not built by default, however by time they will do something else than the Python apps thus misleading users.

Should we keep one/some of them just as an example on how to build C++ apps? There could potentially be applications where routine calls of executables without Python are required (I myself did this years ago) or where apps accompagnied GUI programs as external calls (e.g. ttinv for Ra2dTomo).

[carsten-forty2]

We should keep them but omit there building in the default binary generation.
Do we still ship these binaries in the conda or window package?

[halbmy]

cmake prepares the makefiles in the build directory, but they are on built by calling make apps, which is usually not done in the build process. And they are not shipped with the binaries (neither pygimli nor bert, at least on Windows).

So it does not hurt keeping them but apparently we don't need them either. Anyway we can close the issue to clean up as there is no development and none expected.

[florian-wagner]

I would be in favor of removing both apps and examples folder. We can always go back in the git history if we decide to re-use something.