[Swoop] NCI Investigations

[Bijan Parsia]

So, we did some experiments and determined that parsing is definitely 
not a problem. Indeed, the OWL API is damn fast at parsing. Yay.

We believe that the problem, or at least the current problem, is the 
RDFS reasoner. Which, we estimate, might require >900 megs for NCI :)

We think there's a simple fix (essentially moving from Arrays to 
Bitsets) that will reduce the requirements to closer to 60megs.

For a variety of reasons you might want to adjust the memory given to 
Swoop. You can edit the readme.bat to tweak that (the first parameter, 
stack, doesn't seem to affect anything).

I hope we can do the switch over today, so we might have a version that 
will handle NCI.

Cheers,
Bijan.