Changed the test script to check for different targets. Removed targets that are not supported anymore.

Changed the test script to check for different targets. Removed targets that are not supported anymore.

Removed an unnecessary constant for AKM CCN activation. Added values for AKM CCN activation parameters.
Adjusted the maximum amount of hydrometeors that can vanish in a single RK substep such that in the end of an RK iteration, there is at most no hydrometeor of the given type left and no additional mass is introduced.
Removed some comments.
Added asc600 for the output.
Correlation plots have now better ranges for the x-axis.
get_stats.py was showing the top parameters within two orders of magnitude instead of one.

plot_mse.py for time_plot shows units now on the axis.
Updated README to the current docker version.

Added python_interface for compiling.
model_constants_t and particle_model_constants_t now print to a given std::ostream instead of std::cout.

Fixed a bug where a configuration file does not have any perturbed parameter...

See merge request !31

# Conflicts:
#	.idea/deployment.xml
#	CMakeLists.txt
#	exe_scripts/james_ens_all.job
#	include/microphysics/user_functions.h
#	include/types/checkpoint_t.h
#	include/types/model_constants_t.h
#	include/types/output_handle_t.h
#	include/types/particle_model_constants_t.h
#	include/types/segment_t.h
#	scripts/get_stats.py
#	scripts/parameters.py
#	scripts/test_output.py
#	src/CMakeLists.txt
#	src/microphysics/trajectories.cpp
#	src/types/checkpoint_t.cpp
#	src/types/input_parameters_t.cpp
#	src/types/model_constants_t.cpp
#	src/types/netcdf_reader_t.cpp
#	src/types/output_handle_t.cpp
#	src/types/param_t.cpp
#	src/types/particle_model_constants_t.cpp
#	src/types/segment_t.cpp

Added plots for raw data from a sensitivity simulation. Changed imports to use tqdm instead of progressbar if progressbar is not available.

Fixed a bug for create_train_set, where some gradients are written to the wrong destination or ignored entirely. Added plotting routines for the training data.
Some bands of gradients/predicted deviations still occur. This might be another bug.

Fixed a bug from the bachel_full.json configuration with a wrong parameter name.

Fixed a bug where gradients have duplicate dimension "trajectory" for simulation mode create_train_set.

Using the simulation mode create_train_set now writes the gradients without any scaling, i.e., 10% scale is the default value for all other modes.

Removed -Werror since a user may have installed a version of openMPI (or any other library) which has warnings.

Code can now be compiled with pedantic errors. Can now create a train set (includes a bugfix with optimization level 2 or above). Removed old comments and code.
Increased efficiency for evaluating gradients.
TODO: Check if the output for gradients is still valid. Check if the perturbed output is still valid.

Fixed a bug and now the names of parameters are written to the NetCDF-file. There is still a memory error.

Fixed a bug where output_buffer in output_handle_t.h was not large enough. There is still an error with MPI_ABORT (-57) in nc_put_vara() in output_handle_t.cpp line 1429.

Fixed a bug where compression would not take different simulation modes and their different variables into account.