Not obvious that all DnaComponents must be explicitly added to a collection

[Juul]

The xml format uses a flat list of all DnaComponents and then SequenceAnnotations refer back to that list, which makes sense in order to avoid repeating the same data, but it doesn't mirror the data model, and it is not obvious that you have to add all of the DnaComponents referred to in your SequenceAnnotations to your Collection manually. Right now, if you don't do this, the SequenceAnnotations that have subComponents not listed in the collection will only refer to the Component by URI, and upon deserialization you will realize that you have lost all other information about that Component.

I don't fell that the serializer should work like this. It should find all unique DnaComponents, whether they are in the Collection directly or only included as SubComponents, and it should compile a flat list of all of these, and then refer back from the SequenceAnnotations as it is doing now.