Structured configuration/metadata file via CLI

[vinjana]

A structured configuration file (XML, JSON, YAML) should be used instead of the analysis and dataset specifications.

E.g.

roddy.sh rerun $metadataJson [remainingParameters]

The metadata JSON/YAML/XML would then contain datasets. This does not exclude that optionally parameters could filter to specific datasets/cohorts.

The format of the metadataFile should be so comprehensive as to allow specification of all thinkable configurations, comparisons (like cohorts, supercohorts, linear regression models, whatever). Obviously, because we cannot think of them all, maybe the plugin codes would be place to interpret specific subsections of the metadataFile). The object representation of the full metadata file or only sections of it, would then be givin to the workflow object for further interpretation.