Justify the existence of LongPrimitive, SymbolPrimitive and BoolPrimitive in their kdoc

[dlurton]

This issue is example number 8,274,783 why good documentation is needed: Hours after creating this issue, I remembered why LongPrimitive, SymbolPrimitive, etc, exist: they need to inherit from DomainNode which allows for the generic visitor functions like visitDomainNode, and walkDomainNode, etc, which are invoked for every node and leaf in a tree.

Thus, this factoid should be added to the documentation of these types.

Original issue text follows.

We should consider deleting LongPrimitive, SymbolPrimitive and BoolPrimitive and instead use the IonElement equivalent of these: IntElement, SymbolElement and BoolElement.
The existence of these classes doesn't seem terribly justified given that their IonElement equivalents already encompass all of the same functionality. If this were completed then: all we'd need to do to complete #46 is add a few tests, and it would significantly simplify adding support to all of the other Ion types to Kotlin target.
If we discover a better reason than described in the kdoc of these types, and decide instead not do this, then we should include that justification in the explanation given in the kdoc of these classes.