Adding useNA handling

[jimvine]

This edit implements a useNA option, defaults to FALSE to retain existing behaviour, but providing additional functionality optionally.

[gregfrog]

Jim, I don't know if you are still interested in this, but I'd like to discuss the issue since I am working on the package. It appears to me that the current behaviour of the package is to return nulls for fields without values, which is a problem for aggregation so NA seems like a useful way of getting a handle on those values. These nulls can be mixed with non-null values within the same returned list, does your proposed change deal with that somehow? It looks to me like it only handles the case where the input UA is NA.

I wonder whether a zero length string is a more idomatic value for those currently null values.. The empty values that are returned are not missing in the sense of the value being randomly not provided, but they are actually a value in the same way that a non-blank value is. I'd be inclined to replace the individual null fields with zero length strings and have all NA values as the response to an invalid (NA, null or blank) UA.