parallel search feature requests for Linux

[machth]

1. separate memory and cpu setting for library_prediction, mbr_library and lfq from first_search and second_search.
   -> add default search parameters of cpus_mbr and mem_mbr
   -> positive effect: first and second search can be better distributed over multiple nodes while the other can and need to be run on one node with the highest possible cores and memory requirements.
   e.g.
   Default search parameters
   nnodes=100
   ntasks_per_node=1
   cpus=20
   mem='240G'

new:
cpus_mbr=40
mem_mbr='480G'

2. it should be possible to load in .tsv files -> could also be an alphadia or alphabase problem?

3. Log file (dist_AD-slurm.out) should report in the end that the pipeline was successfully run
   -> currently it ends with "Submitted batch job" for the LFQ step.

4. Include timing lock and calculation in the log file (dist_AD-slurm.out)
   -> include the start time at the beginning and the end time in the end. Additionally calculate the time how long the pipeline needed. Would be also interesting for each separate step of library_prediction, first_search, mbr_library, second_search and lfq.