Decide where to retrieve files from, when replicas exist

[HerrHorizontal]

On systems with many GRID storages, replicas of the same datasets might exist. A job, requiring some files of this dataset, needs to decide where to retrieve them from, when multiple options exist. This also has an implication on caching the file, since the data could be cached from multiple sources. The logic for the cache might be different from the job read action.

[HerrHorizontal]

This might get obsolete with the inclusion of an XRootDService in WRENCH, since this service would take care of providing the files using its own logic.