Move self_api.py and peer_api.py into an api subpackage

Split serializers.py into serializers/base.py and serializers/peer.py

This also creates the AssessmentWorkflowStep model to track all
steps for new workflows. Some problems with it now:

* a lot of broken tests
* some display oddities on the peer step during (peer, self) workflows
* have not tested to see if there are concurrency issues with step
  model creation under load.

This introduces a few features:
* INFO level logging on common events like creating or fetching
  submissions and assessments.
* Parallel event logging for the same.
* Creates a local logs/ directory for development.

For the most part, we rely on XBlock's publish() mechanism to emit
event data. However, because scoring is done at the API layer in a way
that is not directly visible to the OpenAssesmentBlock, we're logging
that event separately with a hacky solution that dynamically loads an
emit function based on a configuration value. This is a stopgap
measure until we can start using the edX analytics API (which is still
in testing).

TIM-260, TIM-378

Add UI styling for course staff, including collapse/expand.
Cache student item when retrieving submission and student item.
Protect against invalid cache key exceptions.
Update test suite to clear cache between tests.
Add composite index for AssessmentWorkflow student item fields

Shift the creation of the peer workflow to happen when overall
workflow is created, instead of implicitly creating it if it
doesn't exist while checking for workflow status. The problem was
that the read check happens in multiple places and AJAX requests
were hitting it at the same time. Because repeatable-read mode
prevents them from seeing each others work, a couple of threads
would find that the peer workflow did not exist and would try to
create it. This cause integrity errors because those constraints
are enforced, even if those rows aren't available to other
processes.

This should have been fixed no matter what, but it's important to
note that Django should never be run under MySQL's default
repeatable-read mode. Use read-committed. For legacy reasons,
edX's own servers are misconfigured in this respect.

[TIM-262]

Wiring all the peer queue into place

Updating based on rebase to fix tests

Implement self assessment api

[TIM-202]