Add support for running BEELINE on multiple datasets specified implicitly

[tmmurali]

At the moment, if we want to execute BEELINE on multiple datasets, e.g., 10 different synthetic datasets created by BoolODE for a GRN, we have to explicitly specific each directory containing a dataset using the datasets keyword in a config.yaml file. This step can become painful as the number of datasets becomes large.

BEELINE also looks for the dataset_dir keyword in the config.yaml file. So, if this keyword is present but datasets is absent, then BEELINE's config parser should examine all subdirectories of dataset_dir and add them to the internal variables that record which directories contain the datasets to be processed.

It will also be nice if the code can examine whether each subdirectory contains an expression file, a pseudo time file, and a reference network with the accepted names.