Add 1.3.x to valid versions in EC2 script and update default Spark version

[bclune]

Minor changes based off of spark_ec2.py shipping with Spark 1.3.1 from https://spark.apache.org/

[AmplabJenkins]

Can one of the admins verify this patch?

[bclune]

My mistake, it looks like the script doesn't like building 1.3.1 from source without some legwork. Closing for now.

[srowen]

Hm, though in principle I'm sure you're correct that this should be added.
Have you described what the problems were separately on the list?

[bclune]

@srowen I agree, the ec2 script should definitely support 1.3.1. As for problems with the 1.3.1 build, it is likely user error on my part (though 1.2.0 works fine).

I didn't see too many significant differences between the script in the official 1.3.1 release and the script here, aside from improvements in the master branch (particularly the --private-ips flag, which is missing from the official 1.3.1 release), but the scripts are different---I can't seem to find anything in git that matches the official release's script.

How is the official release's spark-ec2 script being maintained?

[srowen]

I'm not the best person to ask but it should just be this script and released with releases.

[bclune]

Ah, it looks like the v1.3.1 tag in git has the correct versions, but those changes haven't been backported to master. Theoretically this PR should be fine but I'm not sure if there's anything else that needs to be backported, so I'll defer to someone with more experience in the codebase.

[srowen]

Oh I see. @pwendell looks like you add these version updates with releases. Would it be good for me to cherry-pick those into master or would that interfere with your sub-process here? I'd guess master/1.4 could show everything through 1.4.0