Replace pbkdf2 with bcrypt

[Eriner]

Description:

bcrypt isn't inherently better than pbkdf2, but everything "just works"
out of the box.

• the hash verification routine now only computes one hash per call
• a per-user salt is built into the hash as opposed to the current
  global salt
• bcrypt.checkpw() is immune to timing attacks regardless of input
• hash strength is a function of real time benchmarks and a
  "difficulty" level, meaning we won't have to ever update the iteration
  count

Note that I haven't tested this code myself. Also, this won't be a clean upgrade for current "beta" users, not sure how that should be handled.

Example entry for configuration.yaml (if applicable):

auth_providers:
  - type: homeassistant

Checklist:

• The code change is tested and works locally.
• Local tests pass with tox. Your PR cannot be merged unless tests pass

If user exposed functionality or configuration variables are added/changed:

• Documentation added/updated in home-assistant.github.io

If the code communicates with devices, web services, or third-party tools:

• New dependencies have been added to the REQUIREMENTS variable (example).
• New dependencies are only imported inside functions that use them (example).
• New or updated dependencies have been added to requirements_all.txt by running script/gen_requirements_all.py.
• New files were added to .coveragerc.

If the code does not interact with devices:

• Tests have been added to verify that the new code works.

[balloob]

So one downside of using bcrypt is that it requires an external dependency. They are providing wheels for Windows, Mac and Linux, so that's good:

For Python 3.7 there are currently only wheels for Windows. Windows users have the most difficulty building stuff, so that's good.

Because it's part of the PYCA org, I think I would trust it to remain maintained.

[balloob]

We'll need a migration path. We can only do the migration while a user is logging in. Verify with old pattern -> if ok, migrate to new salt.

[Eriner]

added

[Eriner]

Added upgrade path, though tests are failing because I'm a shitty python dev. From what I can tell in the upgrade test, I need to set the generated password in the underlying structure but I'm not sure how to do that from the test; asking someone here is faster than me spending time trying to figure it out. Also, there are probably issues with encoding/types somewhere. I'm doing my best but would appreciate some help with this one for the sake of time.

[balloob]

Fixed the test. Also made a small change to the migration code. It's 👍 from me but let me know if you agree.

[Eriner]

Glad I wasn't too far off. Now that this is on the right track, give me another day or so of staring at it to make sure that I haven't made any silly mistakes.

[balloob]

Added one extra test to make sure that incorrect legacy passwords still blow up.

[Eriner]

Can we plan to remove the upgrade path and legacy hashing after some period of time? Any unnecessary complexity in authentication and authorization logic makes me uncomfortable.

[balloob]

Yeah, I would say in ~2 months

[balloob]

Going to make this part of 0.77. Because this change is forward compatible, but not backwards compatible, I don't want it to launch any later than 77, which will enable the auth system by default.

[awarecan]

Need add a note somewhere, if user rollback from 0.77 to prior verison, he need clean .storage/auth*.* and .storage/hassio

[Eriner]

Yes, LGTM. Thanks @balloob!