As Matei described: "We're using Jetty to run an HTTP server, not to embed Spark
in a webapp"

Use explicit asInstanceOf instead of misleading unchecked pattern matching.

Upgrade some of the dependencies while at it.

Also enable -unchecked warnings in SBT build file.

Fix deprecations when compiled with Scala 2.8.1

tabulate used if indexed used by function and fill otherwise.

thus clean up the various singletons (RDDCache, MapOutputTracker, etc).
This isn't perfect yet (ideally we shouldn't use singleton objects at
all) but we can fix that later.

Bagel: Large-scale graph processing on Spark