Clarify data access expectations in 'Large Datasets' chapter

[gremau]

The "Large Data Sets" Chapter of the Data Package Design For Special Cases doc needs to clarify the way data submitters should specify access to an offline dataset (contacts should include any person/org that has physical copies of data), and strongly recommend that the publisher (EDI, for example) have a physical copy of the dataset. The way it is written could be interpreted to mean that one could publish an EML dataset with offline data entities to a repository without also making the data physically available there. While thats technically possible, it creates problems for a repository if data availability changes at the non-repository distribution points.