keep levels of IDs consistent throughout

[atn38]

the various functions as conceived currently take the whole JSONLD representation of EML (via emld) and convert them into tables

these tables have varying numbers of variables that can be used as ID variables and primary keys downstream

levels are
package/dataset - scope, datasetid, and revision number
entity - entity position
attribute - attribute order
attribute code - code and definition

for example, to uniquely ID a given entity, one needs the first two levels of ID variables, but to do the same for a given attribute, one needs the first three

however the way code is written now, sometimes the ID levels are obtained/generated independently within each function, which can lead to inconsistencies.

perhaps the way to solve that is to nest functions, e.g. get_dataset calls get_entities which calls get_attributes and so on