Update create_adapter_plugins.py

[jtcohen6]

Improvements to create_adapter_plugins.py. Some of these are things we should do now, some are things we should seek to automate for future releases.

Versioning

• Bump default core version to 0.19.0-rc1, add to .bumpversion.cfg
• Create file dbt/adapters/[adapter]/__version__.py, to enable reporting from dbt --version
• Codify our recommended practice: plugin versions should match dbt versions, with additional suffixes as needed. (E.g. here's how dbt-spark does it.) Change default package-version from 0.0.1 to match the core version, and add setup functions to effectuate.

Testing

• Testing: Add scaffolding for test/[adapter].dbtspec, modeled off the postgres spec
• Add scaffolding for dbt/include/[adapter]/sample_profiles.yml, which dbt will use to populate ~/.dbt/profiles.yml if a user installs the plugin and runs dbt init --adapter [adapter] for the first time. (We should make this work if they just run dbt init, too: dbt init --adapter default to the available adapter #2814)
• Should the script accept/expect a connection_keys argument? If we did, we could use that to populate both of the above. Currently, we expect the user to add in these fields after running the scaffolding script.

Future: dbt Cloud

Eventually, we may want community plugin maintainers of community plugins to include:

• "blessed" dockerfiles of runtime environments
• JSONSchema of connection profile spec (same info as above)

This will enable us to support the plugin in dbt Cloud. We're still a ways away from that, but it's worth thinking about now in terms of the scaffolding we provide.