* Ignore all types of coverage files.

  I was getting things like `.coverage.$user`.

* Refactor existing serializers.

  Some serializers had their metadata all the way at the top
  of the file in separate variables, even though no other
  serializer were using those variables.

* Put things into proper places to make it easier to read
  and change, without having to jump up and down a 1000+
  line file.

* Add new serializers and use them in viewsets.

  These are the serializers for all 4 new endpoints,
  for courses, course runs, programs, and their aggregation.

* Add new mixin for detail endpoint.

  The mixin will add `/detail` to the end of whatever viewset
  that uses it.

* Call isort on the codebase.

* Refactor serializer test code.

  This will make it x10 easier to add tests for the
  serializers we introduced in this PR.

* Add test code for serializers.

* Refactor and add tests for viewsets.