biomechanics-hlrs-gebert/M-DTC-Direct-Tensor-Computation: Major bugfixes and performance enhancements.

All results of finished domains can be retrieved while other domains are still in computation of pending. The status of the computations may be tracked via hexdumps of the status file. Things, that are improved or enhanced: Restart is now working. Depending on the current position of the ranks during the computation, some domains may be lost. They have to be computed in a subsequent computation. An additional program crawls the final results and writes them to comma separated value files with tailored suffixes to retrieve the results. Only the main ranks of the worker communicators write the results of the whole domain to the file system while minimizing the number of files by using the PureDat format. The preallocation of the PETSc math objects is now implemented correctly. The deallocation of the PETSc math objects is now implemented correctly. Memory usage tracking for the whole node of the main rank and the worker communicators main ranks is now implemented. Additional scripts improve the post-processing of the memory usage. Some modules and routines have been restructured for better readability. Various tests have been performed to check wall times and compute times of the computations, but with no immediate improvement on the total number of processors used. Therefore, standard settings still are recommended. An additional script resets the result files of user defined basenames. Various bugfixes and enhancements