Read strategies (was MagmaCollection API)

[Brent Pinkney]

Hi Göran,

I have successfuly used Magma to demonstrate how a transparent OODb can be used
to managed a very large connected graph of objects (millions).

The read strategies were instrumental in getting the performance to an acceptable
level (ie. very fast ;) ).

For some of the scenarios we know how deep and/or which attributes to materialise
and could configure this when the session was initialiated.

In another scanarios where the user was defining an ad-hoc search, I was able too
construct the read strategy before executing the magma collection query. (It worked 
quite well but this was just a proof of concept though.

Finally our deliberate combination of Seaside + Magma gave us great performance 
becuase we could combine a MagmaCollectionReader and a WABatchedList. This
combination meant we need only materialise the first 'page'  of objects when
returning the results. This feature is still true with complex queries unless sorting
or unique results are required.

Cheers

Brent