Add test cases - 'does not detect a word as its own anagram' and
'anagrams must use all letters exactly once'

Changed names of tests to be consistent and changed the order
so that similar tests are grouped together. Added more tests
using the canonical data.

- in journey test, remove @Ignores so that test runs exercise all tests.
- increase the verbosity of test logging to aide in seeing progress as
  tests are un@Ignored and solved.
- saving scripts used to make changes to all exercises.  These are the
  start of a library of scripts making such changes easier.
- make Gradle output more CI-friendly.

With exercism/exercism.io@54e1df3, we can now gather exercises into a
subdirectory. By separating the language track files from the
configuration and management files, it is easier to read the project.

- move gradle config under "exercises" too as it would make a mess
  otherwise.

This reverts commit 3fdc2d1b9c8749bee26140c359d93fc3a2681889.

Individual exercises must include a stand-alone build.grade, today.

We can inject plugin, repositories and dependencies directly from
the top-level build.gradle.

This extracts a common version of junit for all projects (and
updates anagram to use assertj, the junit matchers did not work
with 4.12).

An anagram is defined as

1. the same letters
2. rearranged
3. to make a new word

The existing assertion only checked the presence of the expected word,
but did not explicitly exclude the existence of results that match the
input exactly.