Transient alfi

[EvaAnton]

Addition of time derivative and transient solver
See example channel_flow_transient (still issue with convergence with no-slip BC)

[florianwechsung]

Thanks for getting started on this! A couple of general comments to clean things up before I look into the code in detail:

• please don't commit .swp files and if possible, avoid commiting .msh files. Instead commit a makefile that contains the gmsh command used to generate the .msh from the .geo file (only in cases where we need to guarantee exact reproducibility (e.g. for a paper), we commit the .msh file).
• the proposed change to the driver.py file breaks existing steady NS functionality. since the goal of this is to get merged into master, I suggest either trying to rewrite the run_solver function in a way that handles both steady and transient gracefully, or (and maybe that's easier), to just write a separate function run_solver_transient that runs a time dependent simulation.
• Please use this pull request for discussions and comments such as Problem with convergence for no-slip BC on walls (top and bottom boundary), and not the readme

[florianwechsung]

As discussed on zoom, I suggest you try

python3 channel.py --nref-start 1 --nref-end 4 --baseN 16 --paraview True --discretisation pkp0 --mh uniform --k 2 --solver-type lu --stabilisation-type none --patch star --restriction

and also push the version of the example that works, with a description what needs to be changed to cause the checkerboard pattern.