Add new language/translation of fields for datasets.

[ross]

Datasets should have a field that specifies the language used in the dataset, this should be a two character ISO 639-1 code (see  http://en.wikipedia.org/wiki/List_of_ISO_639-1_codes for list) . Also make sure all tests and migrations are resolved. [2d]

Determine the best approach to having 'translation_of' references to show which dataset the current one is a translation of. This may be a translation of field, or a 'copied from' field or another form of package relationship. [2d]

[ross]

Some translations are currently done with full locales (i.e. pt_BR) but we don't also have pt_PT and so where I would normally default to the language if there are only one translation I am unsure what to do in this case.

Currently supports the short language code or the locale code, but this will need to be resolved before we address #1653

[ross]

Branch exists for #1650

[ross]

Probably not going to do this for ECP as John found a cleaner way of doing it.

[ross]

Closed as unnecessary now.