ssh commands need the -t argument repeated twice if there is no local
tty, e.g. if the process running spark-ec2 uses nohup and the parent
process exits.

Under unknown, but occasional, circumstances, reservation.groups is empty
despite reservation.instances each having groups. This means that the
spark_ec2 get_existing_clusters() method would fail to find any instances.
To fix it, we simply use the instances' groups as the source of truth.

Note that this is actually just a revival of PR #827, now that the issue
has been reproduced.

Right now it seems like something has gone wrong when this message is printed out.
Instead, this is a normal condition. So I changed the message a bit.

- Use SPARK_PUBLIC_DNS environment variable if set (for EC2)
- Use a non-ephemeral port (3030 instead of 33000) by default
- Updated test to use non-ephemeral port too

Consistently invoke bash with /usr/bin/env bash in scripts to make code more portable (JIRA Ticket SPARK-817)