allow public reading of S3 buckets

[cpcloud]

right now boto tries to authenticate and therefore public buckets aren't readable, but you can read them with urllib2.urlopen(url)

example:

this fails if you don't have your credentials set up

In [12]: df = pd.read_csv('s3://nypug/tips.csv')

but this works:

In [15]: url = 'https://s3.amazonaws.com/nypug/tips.csv'

In [16]: pd.read_csv(url)

kind of trivial but would be nice to be consistent