Figure out when we should be parsing symbols from archive symbol table vs. the internal files

[Tom Read Cutting]

This is currently done it quite an ad-hoc fashion. Mainly to do what is needed to get the tests parsing. So this needs to be properly handled and documents.