Parallelize S3 pull/push files + various bug fixes, take 2.

[didip]

NOTE: I am closing PR #10 and replace it with this one to fix import paths so the PR can be merged as-is.

Original Message:

This is PR is essentially a merge of changes between 2 level of forks (aldrinleal and New Relic).

The changelog of what we, New Relic, made:

• Use go fmt on all files.
• Use the official fsouza/go-dockerclient
• Set GOMAXPROCS to consume all available CPU.
• Pulling files from S3 and pushing to S3 are parallelized using fixed pool of goroutines, currently hard-coded at 50.
• Removed the ability to set S3 prefix path. It makes debugging image name difficult.
• Logging functionality is now standarized to using fmt library.
• Added various code cleanup and test fixes.

[relistan]

FYI @lachie, we've integrated Dogestry support into Centurion: https://github.com/newrelic/centurion . It's currently linked to our fork, but if this is merged we'll point it here.

[lachie]

Hey @relistan I'm happy for you guys to take over dogestry.
I simply don't have the time to give it the love it deserves and I'm really happy that you do.
I just created the dogestry org and added you and @didip as owners. How about we move it there and thus remove the fork name ambiguity?
Let me know if you want to chat some more outside PR messages :)

[didip]

I think that sounds great. Thanks for helping us out @lachie!

[relistan]

Sounds good! Thanks :)