Remove mem per cpu and adding nestcheck test

In this MR the nestcheck setting is added to the dynesty sampler. If it is nestcheck = True, a pickle file is created for each run in the output directory storing a nestcheck.data_processing.process_dynesty_run object, (nestcheck needs to be installed). That object can be used then in the post processing in order to compute the implementation and bootstrap errors explained in “Sampling Errors In Nested Sampling Parameter Estimation” (Higson et al. 2018).

There is also a modification in the --mem-per-cpu setting. The default value is changed to None in order to avoid conflicts with some clusters that do not allow to specify the memory per cpu in the slurm scripts.