Credit eligibility/provider refactor

[wedaly]

• Remove m2m relation between credit course and credit providers.
• Separate eligibility and provider APIs into different modules.
• Add API call for retrieving a user's eligibilities.
• Cache credit course list.
• Style the dashboard purchase button.
• Display a link for the credit provider on the dashboard.
• Add analytics events for clicks on the purchase button.
• Expose more credit models to Django admin and add search functionality.

This completes the work for ECOM-1524 (credit display on the dashboard) and makes integration with the course admin tool easier (since the tool won't need to associate courses with providers on the LMS side)

@clintonb please review when you have a moment. This PR looks bigger than it actually is; I split api.py into two modules, so all of those changes are showing up in the diff.

@srpearce There are user-facing messages in this PR that need review. Screenshots below.

Screenshots

• Eligible for credit:
  
• Eligible for credit, deadline is approaching:
  
• Purchased credit, request is pending:
  
• Purchased credit, request is approved:
  
• Purchased credit, request was rejected:
  
• Unexpected error occurred (in this case, something got messed up when fulfilling the user's order):
  

"staff" is the username, and the email will be set to "support@edx.org" in production.

[clintonb]

Any reason these statuses aren't constants? They are used throughout the code.

[clintonb]

Holy moly! 👍

[wedaly]

@srpearce When you get a chance, could you please review the strings in the screenshots?

@clytwynec The Jenkins build is failing in mysterious ways after I rebased onto master. It seems to not be running a new migration to create the social auth tables that was merged recently. Has anyone else run into this? Is there a workaround?

[srpearce]

@wedaly Working on them as we speak.

[srpearce]

Purchase Course Credit

[srpearce]

OK, I've made some suggestions and I have a question inline.

[clytwynec]

@wedaly I haven't heard anything for this issue. Have you tried running the tests locally? Where is the migration you are expecting to be run?

[wedaly]

Further investigation into the test issue: I'm able to reproduce this locally on my branch, but not on master. I think the tables that need to be created were introduced in #8262, but I'm not sure why my branch would cause these failures. Maybe a test ordering issue?

[wedaly]

OK, I think I figured out the test issue. There was a circular import that was causing syncdb to silently fail.