Fixes #87

Without conflicting with #86.

This is almost entirely trivial changes:

- Make dummy methods that should never be called either raise
  NotImplementedError or call self.fail() in tests.
- Standardize the no-coverage pragma
- Add a few pragmas where coverage varies across
  versions/implementations, mostly for the sake of tox (travis should
  hit them all).

* Enable coveralls

Test coverage is not at 100% yet so it should fail this PR.

Also enable pip caching since we are installing more deps.

Update PyPy versions to latest available for speed and because the old
version of PyPy3 can't install newer pip and coverage, etc.

Partially addresses #87

* Even newer pypys

* Despite the list, pypy-5.6 doesn,t work, it must be 5.6.0

* Test sanity cleanups

Partially addresses #87

- Remove manual lists of tests in favor of loadTestsFromName. This
  mostly gets all test runners running the same number of tests.
- Remove `additional_tests`, which was just duplicating the discovered
  tests. No modern test runner seems to need it.
- Change the two compatibility skip decorators to actually use
  unitetest skips, which helps with the reporting.

* Per @tseaver, drop most test_suite functions.

* Port remaining doctest to unittest.

* Make the odd tests work on all supported python versions

Simply the caching of utility-registration data. In addition to