lookahead clarification/tweaks

[JohnPJenkins]

I recently walked through some simulation code with our friends from Japan and came away with a few observations:

1. there's no command-line parameter for lookahead
2. there's no longer any documentation for lookahead, now that the user guide is gone. Our friends were using the default of 0.005 (ns), which naturally led to awful conservative performance.
3. there's no good way to determine what the lookahead should be

Each are pretty simple to address, I would think. 3 can be aided by printing out stats on the time deltas for events (min, avg, max?).

Another thing came to mind while determining what lookahead they should use is whether it is necessary to check all events in tw_event_new for lookahead violations or just events to LPs in different PEs. A lot of models we develop rely heavily on self events with small deltas for control flow or events to "local" components (e.g. a process telling a model-net lp to do some work). We could definitely ramp up the lookahead and hence conservative performance if so, though on the CODES side we would need to make assurances that we're not splitting up LPs within a "group" to different PEs, which can be done easily enough.