Create helper scripts to help analyze what happened with cluster

[mattsfuller]

We need to create some scripts (maybe as a part of presto-admin), which will help to identify issues with presto cluster.

Ideally we should be able to detect:

• long GC pauses based on GC log if enabled
• jvm crashes

It would create timeline of events which happened in given time period:
{code}
presto-admin show-events 24h
2015-01-01 00:00:000 Node 10.10.0.1 started
2015-01-01 01:00:000 Node 10.10.0.2 crashed (Out of memory error)
2015-01-01 02:00:000 Node 10.10.0.3 long STW GC pause (22.003 seconds)
{code}

We should be able to do this based on gc and launcher logs.

[mattsfuller]

This is an extension to the existing collect logs presto-admin command. Basically, it would look through the logs (and maybe also jmx stats) to produce a timeline of what's happening on the cluster.
This seems to me something that would be a fun hackathon project, but not something that's essential to work on right now.