- Added import and export to the testing. As an extension mechanism, it needs to perform well.
- Added 'performance' make target that constructs an 8,000+ task database and runs benchmark queries against it.