Merging Xu's multi-application workload with MPI Sim layer, enabling time-stepped series data output for the MPI traces (single and multiple applications), updating documentation

Moving computation calls at MPI-Sim LP finalize, so that if the application (e.g. lulesh) doesn't have a call to finalize then the communication times don't mess up

Cleaned code. Removed unused collective functionality. Added ability to compute router connections at runtime rather than load from file.

Adding updates for recording simulation's forward and reverse events, documenting adaptive routing congestion sensing conditions for Xu