Refactor database (some additions to @bacongobbler's PR)

[monaka]

This is also WIP same as #112. This is just for organizing some issues. (As you know, my English skill is poor. So it's best to organize issues by code. ;-)

As seeing recent issues posted, I've inspected issues around the recovery mode for a few weeks. It seems there have three issues I encountered.

• Azure Storage dependent (https://github.com/deis/wal-e/issues/13)
• Crashed by psql (in WAL-E) (https://github.com/deis/postgres/issues/123)
• Seems Postgres fails to spawn WAL-E (https://github.com/deis/postgres/issues/127)

They all causes same result in recovery mode. "infinite rebooting loop."

Patches in this PR is mainly for #123. No effect for deis/wal-e#13. May some good side effects to #127 (not sure).

After I applied patches in this PR. Failures by #123 is decreased on my environment.

[deis-admin]

Thanks for the contribution! Please ensure your commits follow our style guide. This code will be tested once a Deis maintainer reviews it.

[deis-bot]

@bacongobbler and @kmala are potential reviewers of this pull request based on my analysis of git blame information. Thanks @monaka!

[kmala]

@monaka i had created an issue to use the upstream wal-e code which should fix most of your azure related issue #129

[monaka]

@kmala Thanks! I'll try the patch if it was shared.

[monaka]

I re-pushed with fixing my typo. That caused controller block.

[bacongobbler]

thanks @monaka! I think I covered the patches here in #112 this morning. Please let me know if there are more changes I need to make in that PR.

[monaka]

I tried re-pushed #112. It looks that patch fixes issue #123. (At least I didn't encounter it in a few times try.) And that is well refactored. So I close this PR. I hope #112 is merged.