Merge lp://qastaging/~kalikiana/u1db-qt/entireContentsInQuery into lp://qastaging/u1db-qt

Store whole document contents in the results and unit test that

The original reason to avoid yielding all contents was in order to conserve memory. I pondered different approaches such as lazily loading the data in the model implementation. Ultimately this would make it more complicated and harder to test, so given the number of bugs I'm focussing on the easy, reliable solution. It can be optimized later.

Caveat: The results don't always have the same structure as the complete documents.
a) Bite the bullet (and maybe swallow it with Synchronizer goodness)
b) Find a heuristic to avoid the difference
c) Introduce a bool entireContents
d) Introduce Query.fields which can be * or desirable fields
e) Accept an extra * as desire for full contents
f) Bump import version to enable the changed behavior