Split documentation to multiple files in docs/ and use markdown

docs/internals/cache.md

@@ -0,0 +1,10 @@
+# Cache
+
+By default, the linter uses a cache to prevent linting the same migration multiple times.
+The default location of the cache on Linux is
+`/home/<username>/.cache/django-migration-linter/<version>/<ldjango-project>_<database_name>.pickle`.
+
+Since the linter uses hashes of the file's content, modifying a migration file will re-run the linter on that migration.
+If you want to run the linter without cache, use the flag `--no-cache`.
+If you want to invalidate the cache, delete the cache folder.
+The cache folder can also be defined manually through the `--cache-path` option.

docs/internals/publishing.md

@@ -0,0 +1,7 @@
+# Publishing the package
+
+A small note on how the linter is usually published to PyPi:
+
+- `python3 setup.py sdist bdist_wheel --universal`
+- `twine upload dist/django_migration_linter-X.Y.Z-py2.py3-none-any.whl dist/django-migration-linter-X.Y.Z.tar.gz`
+

docs/internals/testing.md

@@ -0,0 +1,9 @@
+# Testing
+
+The easiest way to run the tests is to invoke `tox` on the command line.
+
+You will need to install the test requirements, which can be found in the ``setup.py`` file.
+A good way to get started is to install the development version of the linter by doing ``pip install -e .[test]``.
+
+To be able to fully test the linter, you will need both MySQL and PostgreSQL databases running.
+You can either tweak the ``tests/test_project/settings.py`` file to get your DB settings right, or to have databases and users corresponding to the default Travis users.